Intel out of top 5
According to the June 2007 top 500 supercomputer list, Intel is nowhere to be found in the top 5. The fastest Intel powered super computer is ranked #8 with Clovertown. AMD, on the other hand, took #2 and #3 spots with Opteron.
With the death of Itanium and bankruptcy of SGI, Intel simply does not have a viable solution for high performance computing. The 1970s FSB architecture Intel still uses is simply too primitive -- the FSB simply chokes under load.
Once K10 is out, Intel will be totally driven out of the high performance market.
posted by Sharikou, Ph. D. @ 7/08/2007 12:08:00 AM
143 Comments:
Intel more popular in the top 500, AMD becoming less popular:-
# A total of 289 systems (57.8 percent) are now using Intel processors. This is slightly up from six month ago (261 systems, 52.5 percent) and a represents a typical fraction recently seen for Intel chips in the TOP500.
# The AMD Opteron family, which passed the IBM Power processors six month ago, remained the second most common processor family with 105 systems (21 percent) down from 113 systems (22.6 percent) six month ago. 85 systems (17 percent) use IBM Power processors down from 93 systems (18.6 percent) six month ago.
AMD is finished. Pre-fragged CPUs. Pre-fragged GPUs.
Intel beats AMD to market with quad core by ten months and AMD's quad core still can't match Clovertown. Too bad for AMD, Intel has Penryn coming this year. Faster and lower power usage and cheaper to manufacture. AMD is done for.
AMD BK Q2'08.
AMD will lose even more server market share this quarter. Expect ugly results for AMD this month. Intel will report more profit than AMD will in revenue,
Whether you are 1, 10 or 50 is more a matter of how many CPUs you have chosen to put togather. Look at the list and you still find many old NetBust and even some Itanium.
What matters more is what the capability of the individual CPUs can do and in the "Ph"ony "D"octorates own words C2D frags anything that AMD can produce. Damm even the 2GHZ Barcelona is no match for existing C2D let alone the upcoming Penrym and Nehalem.
Oh where oh where are those supposidily benchmarks now? Oh withdrawn by AMD as they are all lies.
AMD BK in 2008, got a broken design, got a fab with low yields, got speed problems, got thermal problems.. got negative cash burn..
Did I mention AMD is going BK?
Sharikou, did they let you out of the playpen again?
You are even more marginalized than ever. Even your buddy Scientia craps all over you with the following:
There is also no doubt that Intel has gained tremendously in servers. In fact, in HPC, while AMD had modest gains, Intel's gains could only be described as ballistic. This is the sharpest gain by Intel in HPC on the Top 500 list since November 2003 when a flood of 32 bit Xeon systems began replacing the RISC systems like Alpha, Pa-RISC, MIPS, and Sparc. In just one year, Xeon HPC system power had doubled and became the dominant source of computing power in June 2004. Today, Intel has duplicated that feat by surpassing IBM's Power with 64 bit Xeon. As we speak, there is more HPC computing power coming from 64 bit Xeon (primarily Woodcrest) systems than any other. Of all the Top 500 HPC systems, 64 bit Xeon systems provide about 20% more computing power than IBM's Power and about double that of Opteron. AMD does have some large HPC systems lined up for later this year but even 30,000 processors for these systems won't put much of a dent in Intel's lead. AMD would need about 70,000 Barcelona's in HPC today to catch Intel.
Here, go feed your face:
http://scientiasblog.blogspot.com/2007/07/on-quality-of-things-ive-said.html
Looking at the top 500 :-
Intel EM64T - 231 systems with 275450 processors achieve a Rmax (sum) of 1792543.
AMD x86_64 - 107 systems with 269850 processors achieve a Rmax (sum) of 940740.
So with almost the same amount of processors Intel achieve an Rmax sum nearly twice the amount of AMD.
(Rmax - Maximal LINPACK performance achieved)
See this AMD presentation:-
http://www.sunmicrosystems.se/virtualisering/pdf/AMD_Quad_Core-Leif_Nordlund.pdf
Look at page 16. There is an IPC improvement estimate of 15% in there. This is why AMD has kept quiet. The scenarios is pretty grim here:-
Barcelona will be on a per core basis 15% higher IPC than K8. The Core Micro architecture is ~20% higher IPC than K8 on a per core basis. (E6600 2.4Ghz is equal to 3Ghz 6000+).
If you allow AMD a small gain for being a native quad core and using HT vs FSB they might be about 5% faster at the same frequency than Clovertown. Penryn will reduce AMD's advantage in that area though, with larger L2 cache and a faster bus.
But at 2Ghz there is no chance against a 3Ghz Clovertown or a 2.93Ghz Tigerton.
Intel Pumpers! Time is coming to stop your lying and for full AMD truth! This supercomputing compilation listing is 100% proof that Intel is ALREADY BK.
Let us look at list here: #8 system is using Dell power. Dell is 100% evil (unless system uses AMD chip, then Dell 100% good). This is evil Intel system and makes for HUGE cancer all over the world. It is so pathetic that the evil FSB can only use less than 10% of the CPUs that super-superior AMD systems use. Also, this system can only clock at the 2.3 Ghz, while super-superior Opterons are clocked far faster. It is 100% truth that the Super Opteron machines use 0 power, while just 1 Intel CPU makes all lights in county go out!
Super-Success companies like Cray have told AMD, why bother with Barcelona? Opteron is so fast today that no evil Intel can ever catch up. But Hector is insisting on making it even faster because he cares about the world, even if Intel intentionally makes exploding laptops. AMD already has 1 Billion Barcelonas at 4 Ghz waiting to go out. The reason it will be released at lower speed is that it is too powerful for many software types and the software will have to re-write to handle all extra Barcelona speed!! AMD is just making life easy for software people who were stuck with evil Intel before!
Last point is major controversy over benchmarks. It is a proven fact that no 'real world' testing is ever accurate unless it is done in scientific way. Since AMD chips are used in real supercomputers, only AMD chips can be scientific so comparing an Intel chip to an AMD chip is pure cheating and makes no sense (like comparing apples with macintoshes)! Only AMD benchmarks can be trusted.
It has been shown here for 100% proof that an Opteron that is simulating a Barcelona super-computer chip is 50% faster than any evil Intel CPU. Some pumpers pretending to be "reasonable" will say that Barcelona is 50% faster than any Intel crap CPU. DO NOT BELIEVE THIS EVIL PRO-INTEL PUMPING! This is only a simulation of Barcelona that is 50% faster. The simulation is far slower than the real chip is!! The real Barcelona is at least 500%!!! faster than Intel CPUs, it was only the Opteron simulating Barcelona that was 50% faster!
It will only take 1 Barcelona to be in top 10 of Supercomputing 500 list.
Intel already BK.
Its been said before, Ph(ake)d only posts what he wants to and fails to realize that people can read.
Well OK most people, as it seems the AMD fanbois that follow him around like some Christ, not only can't read, but can't see, hear or do much of anything except nod heads and claim AMD is the greatest.
Hey Sharidouche care to comment on this:
"There is also no doubt that Intel has gained tremendously in servers. In fact, in HPC, while AMD had modest gains, Intel's gains could only be described as ballistic. "
Of course he won't say a word.
Hopefully he will do us all a favour when Q2 2008 passes with Intel owning even more of the CPU market, and AMD is back to the "has been" and just give up.
theburninator said:
"AMD already has 1 Billion Barcelonas at 4 Ghz waiting to go out."
and
It has been shown here for 100% proof that an Opteron that is simulating a Barcelona super-computer chip is 50% faster than any evil Intel CPU. Some pumpers pretending to be "reasonable" will say that Barcelona is 50% faster than any Intel crap CPU. DO NOT BELIEVE THIS EVIL PRO-INTEL PUMPING! This is only a simulation of Barcelona that is 50% faster. The simulation is far slower than the real chip is!! The real Barcelona is at least 500%!!! faster than Intel CPUs, it was only the Opteron simulating Barcelona that was 50% faster!
Kinda makes our point, doesn't it? AMD pumpers have run out of meds. No money left.
He is being sarcastic......
I say burninator should start his own blog its got the same amount of "facts" found on this blog.....
I wonder what this list says about individual processor performance when the top machine uses 700MHz PowerPC's?
The top machine uses
131,072 processors,
32,768 GB Memory,
1.5 megawatts power,
2,500 square feet
imagine putting a bull into that pottery shop :)
Giant said...
Intel more popular in the top 500, AMD becoming less popular:-
Oh really? Six month trend... what the hell is that worth? How about posting the numbers since 2000? Try that!
Intel beats AMD to market with quad core by ten months
Although technically true, don't compare apples to oranges! No one would ever compare a Mercedes (AMD) to a BMW (Intel) car.
AMD's quad core still can't match Clovertown.
And you must have tested Barcelona yourself for you to know that right!
Penryn [...] and cheaper to manufacture. AMD is done for.
You are a complete idiot! But again, I'll explain a few things here!
Back in the days, when Intel/AMD had a process shrink, costs would go down! Direct result of Moors Law!
But things have changed since then, adding additional costs per processor:
a) Due to increased competition, FAB and processor design depreciation is steeper than ever.
b) Increased cache sizes have significantly increased costs.
c) Yields and processor count per wafer have SIGNIFICANTLY dropped!
I'll explain! When Intel upgraded to 300mm wafers and underwent a die shrink... technically it doubled its processor count, or in other words, cut costs in half (sort of).
But due to:
a) increased die size due to the additional L2 and L3 cache, and
b) lower yields and processor count due to multi-core processors
These have all increased costs.
You would be correct in your statement had Intel/AMD not:
a) produced multi-core processors, and
b) had not put in additional cache memory, and
c) did not compete so aggressively on die shrink time-line.
If you don't believe me, look at Intel's (and AMD's) fiscal statement every year for the next couple of years. Each and every year, their gross margins will continue to drop. The days of high 50% to 60% are over for Intel. Although I will personally still invest in this industry, I can tell you that this industry will NEVER be what it once was in terms of opportunity.
Evil_Merlin said...
"There is also no doubt that Intel has gained tremendously in servers."
Although I have no doubt that Intel has made some gains over AMD in the server segment, I would advise most to read these numbers with a grain of salt.
Because market share is not a percentage of servers sold based on total servers replaced or put in place.
What I'm getting at is that its possible that a few years ago AMD made huge amounts of server sales compared to Intel because the AMD loyalists were active replacing their systems, while Intel loyalists were mostly idle waiting for Intel's next gen.
Now that Intel's next gen is here, most are upgrading their servers, while most AMD loyalists will be waiting for the Barcelona era.
So just looking at current server sales does not indicate sales as a percentage of the installed servers base.
For example, if we pretend that Barcelona is just so so, and we assume that Penryn is amazing. If most people have purchase the current Xeon quads and expect to upgrade to Penryn only in 2 years, all the while everyone in the AMD camp upgrade from dual-Opteron to Barcelona right away, this would give AMD a HUGE market share... but yet very inaccurate overall.
Market share analysis in the short term is meaningless!!!!
a) increased die size due to the additional L2 and L3 cache, and
This is totally false. C2D 4MB Cache at 65nm is 143mm2. A Penryn die with 6mb Cache at 45nm is 107mm2. That's quite a reduction. The dual die solution is clearly superior for quad core at this stage. 2x 143mm2 or 2x 107mm2 makes it much easier to manufacture than a single 283mm2 Barcelona.
By going with with a small die-size product to begin with Intel can work out the yields in the process with something that is fairly easy to make, they have plenty of time to do this. Since Barcelona is coming in August much slower than AMD hoped for, Intel is in no real hurry to release Penryn before late '07/early '08. (latest rumors are late Q4'07 for servers/early Q1'08 for desktop/mobile).
By the time Nehalem is ready for production the yields will be excellent. Intel did the same thing with 65nm, starting with a small die-size product like Core Duo (90mm2).
As for performance Barcelona vs. Clovertown/Penryn see my second post. I used AMD's own IPC estimate as a basis for the comparison.
For die size comparisons, see this:
http://media.corporate-ir.net/media_files/irol/10/101302/20070529GelsingerOshanew.pdf
See slide 14.
Giant: How do you find those PDFs?
Gee, If Intel knew back then when they made that presentation, about AMD's scam, heck they'd make good use of it :D
Jeach!"
"Six month trend... what the hell is that worth? How about posting the numbers since 2000? Try that!"
Here you go. Number of systems with AMD and Intel CPUs in top500 list. I used the data from the last 15 list, starting with the first list that came out in 2000.
As can be seen Intel has always been in the lead of AMD. With AMD 64 things got a bit difficult for Intel. At least in the last list the growth of AMD systems has become into decrease. It should be a bit more interesting when the second list of this year gets released around November.
"Although technically true, don't compare apples to oranges! No one would ever compare a Mercedes (AMD) to a BMW (Intel) car."
Both have four wheels and move, this is all that I care.
"And you must have tested Barcelona yourself for you to know that right!"
Can you prove he is wrong? All indications say that in 1P/2P Barcelona can't match highest speed Clowertowns in most tasks thanks to having not too great IPC and low clocks.
"a) increased die size due to the additional L2 and L3 cache, and"
As was said the die area shrinks considerably with first 45nm CPUs, even with 50% more cache.
That's quite a story, Enrique. I like stories. I like stories about pinatas. In fact, I like everything you have to say.
Com on Ho Ho! You can't go around posting facts, it ruins the fanboy fun!
One interesting note on the newest top 500 is that the first Core 2 based system (called Abe at NCSA in Illinois) in the top 10 has appeared, and core-for-core it's Cloverton CPU's at a measly 2.33Ghz are faster than the much larger and more power hungry Opterons that are running at 2.6Ghz on the #2 machine. Since Core 2 is such a new architecture expect to see a lot more of these in the near future. It takes time & planning to deploy a massive supercomputer which is why the Core 2's are only now appearing in the list even though they have been available for 1 year (same thing happened when Opteron was new).
Just compare the numbers: Doubling the number of cores on the #8 machine would result in a machine that only has 4800 sockets/19200 cores (meaning smaller footprint) and a higher rmax and rpeak than the #2 Opteron configuration. slack_comp_user also has a good point: Supercomputers don't need to have the fastest chips which is why a whole crapload of slower PPC chips is in the #1 slot, and why Abe is only using 2.33Ghz chips instead of faster models.
As for you AMD fanboys who scream about the FSB, I have news for you: IT DOES NOT MATTER The reason is that a modern supercomputer is not actually a single computer at all. Instead it is a cluster of hundreds or thousands of individual nodes. The FSB does not have to scale that high inside of each node (each node is really just a somewhat stripped-down server or blade). What is more important is that the nodes can communicate back and forth very quickly and with low latency. I'm not talking about ethernet you use on a LAN, I'm talking about Myrinet, Infiniband, SCI, etc. One of the more popular of these links is Infiniband which was originally invented by...... Intel (although a bunch of other companies use the technology too and it is standardized). So, when it comes to supercomputing while Opteron is definitely not a slacker, the new Core Architecture has already starting making its presence known, and if Barcelona gets flubbed AMD is in for a world of hurt.
pinatas?
AMD is currently Intel's pinata.
With the death of Itanium...
what happened to the name "itanic"? why doesn`t anybody use this term anymore? Has the itanium bashers gone weary of using the name of a boat that sunk in 2 hours to a processor that is now staying around longer than their will to ridicule it.
It`s nice to realize how silly they all were.
Not to mention, Itanium installs are actually growing, year over year. Yeah its nothing major, but it is a full, true 64bit processor, which is something AMD does not make.
Anyone who claims x64 is a true 64 bit CPU, needs to be shot, run over, and shot again.
The Itanium can (and does) do things AMD's CPU's dream about.
Giant: How do you find those PDFs?
Intel maintains a nice list of it's most recentpresentations here:-
http://phx.corporate-ir.net/phoenix.zhtml?c=101302&p=irol-presentations
I'm not sure if AMD does the same. I saw a link to the AMD presentation on a forum.
Just more (FUD) from the inzel pumpers/plumpers
This comment has been removed by the author.
Ah Oneexpert, good to have you back. So tell me, if Cloverton is just some old P3's glued together, why is it that the Abe system at NCSA is core-for-core faster than any of the Opteron systems in the top 500? Now, don't blow crap at me about how it is only ranked 8th, that's like saying that owning 500 P4's is faster than owning 1 opteron, you need an equal comparison.
So lets take you at your word. 1. Cloverton is a crappy glued-together P3 setup. 2. The statistics from the exact same Top 500 list Sharikou himself trumpets show each crappy core is faster than the fastest Opteron cores in the same list. 3. Conclusion: The P3 is superior to anything AMD has been able to produce.
You see how bad that makes you look? You should be complimenting Intel so that if Barcelona actually wins you can say how great AMD is instead of screaming how you actually think AMD chips are slower than P3s.
As for your C7 nonsense, that is just the time it takes the CPU to switch power states. An old P4 is faster at actually running software than any modern C7, although the C7's are power efficient I will give that to Via (although Via hasn't been extremely successful with them commercially). Oh and one more thing: Be careful about bragging about Via since it can also make your main Golden Goose of AMD look bad too. I mean, AMD's power technology is a joke compared to what Intel can do, witness the notebook market.
And finally, yes I know that you are actually an Intel fanboy who is just spewing nonsense to make the other fanboys look stupid. Your extensive list of Intel CPUs you've purchased (even though you were insulting them) is more than enough evidence to show who you actually prefer.
This comment has been removed by the author.
why is it that the Abe system at NCSA is core-for-core faster than any of the Opteron systems in the top 500?
Faster based on what? Integer? Floating point? That doesn't compare two systems core-for-core.
Try giving me benchmarks in the area of:
a) database trasactions per second
b) web transactions per second
c) virtualized server with over 100 servers running in parallel.
Then we'll see who's on top
Here you go, Jeach.
http://www.sap.com/solutions/benchmark/index.epx
Not an AMD system in the top 50 that I can see.
I can't wait to see how you argue this one!
Jeach.... you make me feel like Chris Tucker from Rush Hour "Can you understand the words that are coming out of my mouth?"
Step 1: Go to the top of this page where Sharikou (you know the guy who runs this blog) posted his inane rant.
Step 2: Actually click on the link going to the newest list of Top 500 supercomputers. Since this may be too hard for you here is the relevant page.
Step 3: Look at the 'Rmax' values for the #2 system (running Opterons at 2.6 Ghz) and the #8 system (running Cloverton at 2.33 Ghz). You can look at the Rpeak scores too, but Rmax is the 'sustained' rate and Rpeak looks even worse for AMD so I'm doing them a favor.
Step 4: Realize that there is this word called 'context' that applies to this discussion. In this case the 'context' is a discussion of Supercomputing applications where the Linpack benchmark for floating-point operations is the standard.
Supercomputing applications do not care about:
1. database transactions;
2. web transactions;
3. 'virtualized server with over 100 solitaire games';
4. Playing Quake 4 (this is not a desktop!)
They never have, and I doubt they ever will.
Step 5: Realize that I am smarter than you and have already anticipated that you are now going to start screaming about how Linpack is somehow not a real benchmark because it isn't showing AMD in the best theoretical light.
I don't really care, it's a valid benchmark that has been in use for a long time, and when AMD was cleaning up 2 years ago it seemed perfectly fine for AMD fanboys to quote then, so it's still perfectly fine to quote now.
Step 6: Do some rudimentary math. Unlike your friend Pezal I'm not going to arrive at a ridiculous conclusion and then fudge the numbers backwards, I'll actually do some (very basic) real math:
Rmax for the #2 Opteron system: 101700
Number of Cores for the #2 system: 23016
Score points per core = 101700 / 23016 == 4.419
Number of sockets for duos: 11508
Rmax for the #8 Core 2 system: 62680
Number of Cores for the #8 system: 9600
Score points per core = 62680 / 9600 == 6.529
Number of Sockets for Quads: 2400
Now let's see what would happen if we double the number of CPUs in the Intel system. This still leaves 3816 extra cores for the Opteron at #2, so it is not a 100% core to core comparison (but it will prove my point). Also, even though Linpack is a very highly parallelized benchmark, I will give a 10% overhead factor to the scaling since in the real world simply doubling the number of nodes will not perfectly double performance:
New Intel system with 19200 cores:
19200 cores * 0.9 efficiency * 6.529 points/core = 112821
New number of sockets: 4800
A score of 112821 beats 101700 by 10.9% while using 3816 fewer cores.
Also, the new system only requires 4800 sockets vs. 11508 sockets (a savings of 6708 sockets).
Step 7: So it looks like even spotting the Opteron an extra 3816 cores at higher clockspeeds (all of which are sucking floorspace and power down) the Intel system is faster.
But wait, Sharikou just gave us this whole Song & Dance number about how power consumption is so important!! Well I got bad news for you, since in addition to the fact that the Intel machines use less power for each core, the very fact that there needs to be 6708 fewer sockets for the Intel machine means that fewer nodes with fewer power supplies and fewer connectors and fewer cooling fans and fewer everything are needed to get the higher performance.
This means: POWER SAVINGS over the AMD system.
Step 8: When you can show us any kind of semi-rational number for Barcelona performance, I'm all ears. I'm not holding my breath though. (and simulations of what a chip AMD will not even sell you this fall might do are not the same thing).
Oh, one more thing.
Please provide a link to ANY info on an AMD system running hundreds of virtual machines.
After you can't find one, you may then come to the realization that it's impossible.
The AMD fanboys ranted about a tiny 7.5m investment in Transmeta by AMD. That's pocket change to a company like Intel. Intel has invested over $200m in VMware.
http://www.theinquirer.net/default.aspx?article=40873
We can put that $7.5M in perspective.
It's how much money AMD loses in six hours.
Or, looking at it another way, the amount of money Intel makes in two hours.
Way to go Hector!
Please AMD, get smart fire that beaners ass.
AMD totally fragged:
http://enthusiast.hardocp.com/article.html?art=MTM1OSwxLCxoZW50aHVzaWFzdA==
For the most part two Radeon HD 2900 XT cards in CrossFire improved performance over a single card configuration. We saw a very large improvement in S.T.A.L.K.E.R. That still only barely came close to one single BFGTech 8800 GTS OC2 640MB video card’s performance! With ATI’s 2900 XT and S.T.A.L.K.E.R, you’d have to spend $800 to get close to the same performance that you can get for a $400 NVIDIA card.
AMD BK Q2'08.
This comment has been removed by the author.
oneexpert wrote:
"Transmeta accuses Intel of patent infringement..."
Superbly written rhetoric. Meets all expectations of a true green AMD fanboi. Keep up the good work, son. Hector will be proud.
Transmeta accuses Intel of patent infringement
By Tony Smith [More by this author]
Intel has filed a lawsuit against chip designer Transmeta, alleging that the company infringes on seven of its patents:
http://marketwatch-cnet.com.com/Intel+countersues+Transmeta/2100-1006_3-6149378.html
If AMD did buy Transmeta the litigation would stop. Intel and AMD have a broad cross-licencing agreement that says they can use each others IP. Part of this complex agreement also includes AMD's right to develop CPUs based on the x86 architecture.
giant let me add this review too:
Crossfire HD 2900 XT vs. 8800 Ultra
What helps Crossfire is the fact that the P35 is such a fantastic board and only currently supports Crossfire as a multi-card solution. So if you want an excellent motherboard that is going to give your CPU huge overclocks, along with the graphics performance to go with it, Crossfire is the only solution. When compared to Nvidia’s SLI arrangement, Crossfire generally manages to scale better as well, everything saw a performance increase and at 2560 x 1600 there were some pretty significant ones.
The pros are there, performance, performance and performance. What about the cons though? Well you’re going to need a decent power supply, but then you’re going to want a decent one for an 8800 Ultra as well.
Whoops looks like today AMD's processors are in a price tailspin down to nothing.
So lets see how the AMD fanboi's react to this one.
Remember just a post ago on this fuckhole of a blog, it was claimed Intel lowering the prices on its products was a very bad thing...
Sounds like AMD really has its back against the wall.
Is it just me or that Crossfire system was compared against single GTX/Ultra?
Well according to hardopc one 8800GTX is better than two 2900XT, but according to Tweak town one 8800Ultra cannot beat two 2900XT.
Well who to believe?
aguia
"Well according to hardopc one 8800GTX is better than two 2900XT, but according to Tweak town one 8800Ultra cannot beat two 2900XT."
Main reason for that is that Tweaktown didn't use AA in games. XT's have awful AA performance, especially with higher resolutions.
Had they used AA and more games with bad dual GPU scaling things would have been a bit more clear.
With roughly equal performance between XT CF and single Ultra or GTX I'd definitely go with the single GPU as you can be pretty sure that with badly supported games the second card doesn't sit idle.
Also has AMD already fixed the problems with Stalker?
The head ATI just abandoned ship!
This comment has been removed by the author.
giant
Look at page 16. There is an IPC improvement estimate of 15% in there. This is why AMD has kept quiet. The scenarios is pretty grim here:-
Barcelona will be on a per core basis 15% higher IPC than K8. The Core Micro architecture is ~20% higher IPC than K8 on a per core basis. (E6600 2.4Ghz is equal to 3Ghz 6000+).
You're such a smart guy, as only Intelers could be.
If AMD's projected IPC gains per core equal to 15%, this means that one Barcelona CPU has about 130% improvement over dual-core Opteron.
Clovertown does not have such level of improvement over Woodcrest, bud...
http://www.theinquirer.net/default.aspx?article=35723
As of Intel's very own presentation their double-chesseburger tops out at 75% performance improvement over their single cheeseburger.
If AMD's projected IPC gains per core equal to 15%, this means that one Barcelona CPU has about 130% improvement over dual-core Opteron.
You wish. Well, AMD wishes as well. You're expecting 100% scaling from dual core to quad core? Not happening sorry. Did you also forget that K10 is clocked to a mere 2Ghz vs. 3Ghz for K8? K8 has 50% more clockspeed than K10.
The 75% scaling Clovertown offers is quite good for an MCM. Clearly, getting product to market ten months before the competition and with plenty of CPUs to satisfy demand is a far better choice than insisting on native quad core and having tons of problems like AMD.
The people at TweakTown are fools. They benchmark the fastest video cards money can buy without AA/AF. Who is going to spend $400 or more on a video card and not use AA/AF?
The 2900 chokes when AA/AF are enabled. Clearly, TweakTown are paid AMD pumpers.
David Orton quitting AMD. He knows Hector Ruiz and Co. are driving DAAMIT into the ground and he wants nothing to do with it. I wouldn't be surprised if he's offered a job at Nvidia.
Finally, brand new Core 2 Extreme mobile CPU: http://www.matbe.com/articles/lire/389/core-2-x7800-de-l--extreme-dans-votre-portable/
AMD Turdion 64 fragged.
AMD BK Q2'08.
bud
Give up off your innuendo.
I wouldn't be surprised if he's offered a job at Nvidia.
I betting that after his non-compete expires, he winds up at Intel.
blah! blah! blah!
DOH! DOH! DOH!
blah! blah! blah!
DOH! DOH! DOH!
Wow, that's an inteligent post.
blah! blah! blah!
DOH! DOH! DOH!
Yeah. These AMD fanboys don't even have a modicum of pro-AMD news to post. Tough times to be an AMD fanboy for sure.
Intel and Nvidia; Clearly the smarter choice!
P.S. AMD stock is more than $10 below Intel stock!
AMD BK Q2'08.
Barcelona can't touch Penryn:-
http://www.fudzilla.com/index.php?option=com_content&task=view&id=1855&Itemid=1
I heard that Pat Gelsinger has ordered the Penryn and Nehalen troops to commence their attack early. The sharp drop of prices of Core 2 Duo is just the preparation for much faster chips to be launched soon. The Penryn will open a performance gap that will never be filled -- as AMD BKs in Q2'08.
Me wonders if Sharikou and the rest of the dolts are going to start claiming Fudzilla is now an Intel pumper...
This comment has been removed by the author.
Bubba said...
I can't wait to see how you argue this one!
Can you pin-point where an Intel Xeon can beat an AMD on a core-per-core? I've looked, but all I can see is the architecturally defunct Itanium.
I've looked and all 8-core AMD's are ABOVE all the 8-core Xeons. Plus the AMD's have a considerably lower frequency!!!
Chuckula said...
Industry standard?
Vehicles are measured in HP... that's THEIR industry standard! But you've still got to consider the weight-to-HP ratio! You've still got to realize that even though your car is made to go 200 Mph, it can't do an average curve at over 100 Mph without taking the ditch.
You go ahead and buy your servers on the basis of Rmax. When your servers crawl to the ground because they can't do average work... well we'll sell you some more Intel crap!
Companies, agencies, schools or whatever that ONLY need integer or floating point power and nothing else are VERY, VERY limited. These companies wouldn't even go with a standard x86 system, they'd buy something that has a vector-processor instead for true SIMD and MIMD.
And just to add my two cents worth, I don't believe that most companies that own less than a dozen servers (quads or duals) really care about power efficiency. And if they do, its a really low priority. Companies like Google on the other does!
Bubba said...
Please provide a link to ANY info on an AMD system running hundreds of virtual machines.
Read my post again for I said no such thing! Just in case it was an honest mistake or I used the wrong wording I'll repeat myself.
I said running virtual servers, but never specified how many. Could be two, could be 5 or it could be 10 I don't know.
But I did say 100 servers, meaning: ftp, httpd, sshd, timed, mysql, oracle, msql, tomcat, etc, etc, etc, up to 100 of them.
Evil_Merlin said...
Whoops looks like today AMD's processors are in a price tailspin down to nothing.
So lets see how the AMD fanboi's react to this one.
You actually think most of us cares? We have better things to do than watch prices :)
Remember just a post ago on this fuckhole of a blog
Why don't you leave? No one will miss you! Why do you keep coming back if you hate this place so much?
I'm starting to understand the pattern here:
You hate this blog so much, but yet you keep coming back for more.
You hate Intel so much that you keep buying more?
Bubba said...
I betting that after his non-compete expires, he winds up at Intel.
Non-competes are non-binding!
There is no court in the world that is going to prevent a guy from earning a living just because he has specialized in a specific domain. Courts are lenient towards that.
If he steals IP or shares any IP, even from his memory, then they could sue on that basis.
This comment has been removed by the author.
The Tweaktown graphics test said that it was $700 8800Ultra v $800 Crossfire Ati, but newegg quotes $599 for one 8800U and $800 for the Crossfire setup. So that is a difference of 33%.
The difference is actually more than that because for the 8800U you can use a 650-750w range power supply but for Crossfire you definitely 900w to 1000w for your comfort zone, so the difference is likely 40% more cost for the Crossfire solution, plus bigger power bills.
Another problem is that you should be able to overclock the 8800U to about 650-670Mhz or so, or an increase of 6-8%, something it seems impossible to do with the Crossfire setup. Add 7% to all the 8800U graphs and it is not too far behind, certainly not 40% behind.
Those extra few frames are very expensive, and like someone mentioned with SLI or Crossfire you really should be running AA with that sort of setup and the HD 2900XT has to do AA in the shaders, not hardware in the ROPS so this gives a hit.
evil_merlin
Me wonders if Sharikou and the rest of the dolts are going to start claiming Fudzilla is now an Intel pumper...
ya know, there is a reason why Sharikou only quotes Newegg users now..
oneexpert
"In the absence of software that is highly optimized specifically for the new architecture of Intel Xeon multi-core processors, AMD holds the advantage in both performance per dollar and performance per watt of power expended."
You do know that all that is needed from programs to work well on MCM CPUs is to reduce communications between threads? Basically it needs decent multithreading, something like in PovRay is quite nice. Stuff like older version of Cinebench is an excellent example of awful multithreading, I'm not sure about the scaling with the newer versions. Come on, <1.5x scaling with doubling the core count? Even a child with couple of years of programming experience can do a lot better. It's just as if the program was specifically made to scale badly.
jeach!
"Can you pin-point where an Intel Xeon can beat an AMD on a core-per-core? I've looked, but all I can see is the architecturally defunct Itanium."
Isn't that interesting, a dead CPU beating the greatest of the great ones!
"Vehicles are measured in HP... that's THEIR industry standard!"
US or UK HP? Also the standard is actually kW.
"You go ahead and buy your servers on the basis of Rmax. When your servers crawl to the ground because they can't do average work... well we'll sell you some more Intel crap!"
Rmax is the load the servers can handle at maximum load. Average load is less than that and servers can handle it easiliy.
"And just to add my two cents worth, I don't believe that most companies that own less than a dozen servers (quads or duals) really care about power efficiency."
I take you have no idea how much it costs to cool those servers ...
"But I did say 100 servers, meaning: ftp, httpd, sshd, timed, mysql, oracle, msql, tomcat, etc, etc, etc, up to 100 of them."
Great, now show us one AMD system doing that.
oneexpert
"An Intel Xeon-based server currently costs about 25 percent more money to build than its AMD Opteron counterpart."
Could you link to those servers you are comparing?
"Intel dual-core servers are more expensive to build than those based on AMD"
What about Intel 2P quadcore servers vs whatever AMD has to offer?
"Also, the increased power consumption, cost and heat dissipation of the FB-DIMMs do not bode well for Intel dual-core server system efficiency when compared with AMD Opteron-based machines."
I take you are not aware of the low-power FBDIMMs that take as much power as regular DDR2
"Evidence of this is visible in the increased market share that AMD Opteron servers have been achieving since AMD's entry into the server market."
Evidence? I'd like to see that, can you link to that evidence, please?
"Experts dont lie."
I know about oneexpert who does, from time to time :P
andyw35
"The difference is actually more than that because for the 8800U you can use a 650-750w range power supply"
Wasn't it so that Ultras used a new core revision that required less power than GTX did? I don't think you really need that high power PSU with Ultra.
Wasn't it so that Ultras used a new core revision that required less power than GTX did? I don't think you really need that high power PSU with Ultra.
That's what Nvidia said. Power consumption is fairly similar between the GTX and Ultra. A decent 500W or higher PSU with at least 30A on the 12V rail should be enough for either card.
Go figure an AMD faboi caught in a lie:
"Intel dual-core servers are more expensive to build than those based on AMD"
Funny according to HP the DL380G5 starts at 2049 (with dual core Xeons). The AMD version, the DL385 starts at 2059, or 10 bucks more. Mind you the DL380 even ships with FB-DIMM's which are superior to the DDR2's in the DL385.
Maybe I should price up some quad-core comparisons.
Oh wait! AMD doesn't HAVE a quad core.
AMD will invest approximately $2.6 billion on new equipment for the facility, which president and CEO Hector Ruiz hinted would focus on 32nm chipmaking process technologies.
AMD is going to need to take on more debt. Expect AMD to report another heavy loss this quarter. Expect Intel to report large profits well in excess of $1bn.
Crysis developers are very impressed with Penryn:
http://www.fudzilla.com/index.php?option=com_content&task=view&id=1894&Itemid=1
Real quad core performance. Only Intel has quad core. AMD has nothing but vaporware. Pathetic.
AMD BK Q2'08.
Ho Ho: Good responses, you put it to Jeach & Chong (er.. Oneexpert), I'll just add a couple of points:
Jeach:
You seem to know about as much about processor architecture as you do about cars... which is nothing. Why is it that whenever I or anyone else post actual facts you have to use a non sequitur car analogy?
Since you don't know what that term means, non sequitur means a logical jump to a conclusion that makes no fucking sense. Here's an example: Cars need Gas!! Therefore all AMD CPUs are faster than Intel CPUs!!! I start off with a true yet completely meaningless statement, and then jump to a conclusion that is completely unsupported by any facts... just like everything you post!
If you want to sound smart, try posting something that actually has factual content to back up your points instead of just saying AMD is faster... because somebody somewhere drives in a car. I know the math I did there is at a 4th grade level and is therefore too difficult for you to understand, but there are programs out there that can help you, it's not too late.
Oneexpert:
OK, with you it's just too easy.
You insult Xeons because they apparently have a 'new architecture' that no software will run on..... BUT AT THE SAME TIME YOU SAY XEONS ARE JUST P3'S!! So what the hell is it, are they old P3's or are they new... you can't even tow a consistent line when you insult Intel. Also, if the K10 is so fucking revolutionary, how is software that can't handle a 'new' Xeon going to magically be super-optimized for a K10???? I'm not holding my breath waiting for an answer.
Next for TSMC: So let's get this straight: Intel is evil and hates America because they build 45nm fabs in the US and produce chips in the US. AMD is somehow this perfect company because instead of even trying to build a 45nm fab they are going to outsource all of their chip production to the Taiwanese.
I hope you realize that bragging about the TSMC deal you are really broadcasting to the world the fact you actually think AMD cannot even manufacture its own chips. I know you think you are really hurting Intel, but they are laughing all the way to the bank.
To Both of you Guys: It's not that you are drooling short-bus retards, it's that you take yourselves so seriously. Look at that burninator guy, when he posts it's actually freakin' hilarious because he's so over the top. You guys are just annoying. Even though I'm pretty sure oneexpert is really an Intel fanboy posting stupid stuff on purpose, try to make it amusing oneexpert instead of just annoying!
This comment has been removed by the author.
Chuckula,
Great post. So good, it makes my replying to oneretard pointless!
I've looked and all 8-core AMD's are ABOVE all the 8-core Xeons. Plus the AMD's have a considerably lower frequency!!!
Wow, AMD fanbois are too stupid to click on the sort links at the top of the columns.
So what the hell is it, are they old P3's or are they new... you can't even tow a consistent line when you insult Intel.
I nearly fell out of my chair laughing when I read that one!
Reminds me of when Sharikou claimed a Core 2 Duo is just a Pentium 3 with 128bit SSE!
Intel making further virtulization investments:
Embedded software startup VirtualLogix, Inc. (Sunnyvale, Calif.) announced Wednesday (July 11) it has secured $16 million in Series B financing from investors including Intel Capital.
Chuckula said...
Why is it that whenever I or anyone else post actual facts you have to use a non sequitur car analogy?
I post analogies because its a proven teaching method used by teachers to help the 'slower' ones of the class.
I was just trying to help you... no need to get insulted!
Here's an example: Cars need Gas!! Therefore all AMD CPUs are faster than Intel CPUs!!!
Yep, great example alright! Normally a teacher would lightly pet you on the back in reassurance. I'm pretty sure you know exactly what I'm talking about.
If you don't understand my analogy, then your the type that they would place in special learning. Don't be embarrassed Chukula... we're not all the same.
----
When do you know that Intelers REALLY lost it?
When they start responding to how great their last post was :)
CHUKULA: Great one 'ho ho'... you really put it to him!
HO HO: Thank you 'chukula'... I tried my best. You are great too...
BUBBA: Your SO amazing guys... you complete me!
jeach!
Here's an example: Cars need Gas!! Therefore all AMD CPUs are faster than Intel CPUs!!!
I'm pretty sure you know exactly what I'm talking about.
Yep, I know exactly that you're saying you're the most illogical AMD fanboy on this forum.
Based on the same logic, I can also say, "Human need food, so therefore it can be concluded that AMD currently is losing its innovation drive".
Clearly, David Orton was a smart CEO. He successfully lead Ati against Nvidia and completely turned around the company during his tenure as CEO.
But all that changed when AMD bought Ati and Hector Ruiz, Henri Richard etc. started running things. Clearly, the delays of R600, it's abysmal performance etc. are all because of AMD. Had Ati remained a separate company, Ati would have still been competitive with Nvidia and would have had the huge chipset deal with Intel. Ati was profitable under Orton, now all this has changed because of AMD and Ruiz.
Hector Ruiz is a sham. He is directly responsible for the current state of Both AMD and Ati.
No matter, AMD will BK in Q2'08 and Ruiz will be out on the streets homeless.
AMD BK Q2'08.
This comment has been removed by the author.
This comment has been removed by the author.
Jeez, those AMD cpu sure do suck a lot of power.
I think I'll stick with the Intel quads that are only 12.5 watts per core, instead of the 22.5 watts per core of the AMD chips. And each one of those Intel cores can do more work in a shorter amount of time than an AMD core.
Twice the work, consuming half the power is good for the planet.
Intel stands alone to save our planet.
Intel best energy saving cpus and platforms.
Of course we haven't even begun to talk about Intels really low power chips yet, the ones in the 1 to 5 watt range.
Jeach,
When do you know AMD'ers have really lost it? When they have resort to attacking the people, becuase they can't use facts for their arguments.
So, have you learned how to click on those sorting links yet?
This comment has been removed by the author.
Intel® Core™2 Duo Mobile Processor U7600 available in laptops!
One company stands alone with the energy saving 10 watt dual core mobile cpus...Intel...Intel...Intel
Saves money when you buy it....
Saves money when you run it....
Good performance and it saves our planet.
Lets all switch over to Intel® Core™2 Duo Mobile Processor U7600 the 10 watt planet saver.
Intel stands alone to save our planet.
Intel best energy saving cpus and platforms.
This comment has been removed by the author.
U2500 - Dual Core 0.75 watt.
AMD should be shut down for destroying the planet.
This comment has been removed by the author.
oneexpert
Would you point me to an AMD dual redundancy fault tolerant server?
oneretard, you are off by a factor of two in your power calc, at least according to AMD:
http://www.amd.com/us-en/ConnectivitySolutions/ProductInformation/0,,50_2330_9863_13022%5E13057,00.html
1.8 watts typical.
Oh, and it's a 500Mhz single core, manufactured on 130nm.
So you need at least six of them to perform the same as a single u2500.
oneexpert said...
u2500...vapor ware/intel paper launch no price, no chip, no sale.
HP Compaq nc2400 Ultra-Portable Notebook
sure does look like vapour ware.... not
This comment has been removed by the author.
Enumae said
Would you point me to an AMD dual redundancy fault tolerant server?
What point are you trying to make?
This comment goes back a few articles/post, and is nothing more than a childish jab at oneexpert's "AMD is god" mentality.
This comment has been removed by the author.
It's a shame that they once stood for performance as well :)
At least the AMD X2 BE-2300 and 2350 balance the power consumption of the 6000+ and quadfx.
Where is AMD's quad core?
oneexpert
What is your opinion then of Intel's MoDT platform?
OK oneexpert, since it's ALL CAPS time for you, explain this page from none other than Tom's Hardware: right here
This is a review of the new L2 stepping DESKTOP CPUs. The E6300 even in the 'old' configuration is only using 45 watts under load (I thought you said that was unpossible?), and in the new power-efficient stepping is used LESS than what your amazing AMD uses.
Of course, the E6300 is also a faster CPU so it takes less time to complete tasks, while also using less power, which means it uses less energy. Normally I'd show you some simple math to explain how power and energy are related, but you wouldn't understand it anyway and Jeach would just say that "Cars have Engines! Therefore AMD CPUs are better!"
Suffice it to say, power integrated over time gives energy, so a lower power Intel CPU that takes less time to complete a tax beats the AMD system in 2 ways for energy savings.
Now, I know you're going to call Tom an Intel Fanboy, but if you read to the end of the article he's all full of praise for AMD (even though the Intel systems use the lowest energy and perform faster). If anything Tom is pro-AMD, it's just that the facts speak louder than your useless words.
Hey tard, I guess these five merom boards aren't desktop then, huh?
http://usa.aopen.com/Products.aspx?id=52
Or this one?
http://www.gigabyte.com.tw/Products/Motherboard/Products_List.aspx?VenderType=Intel&CPUType=socket+m+(478+pin)
How many more examples would you like me to provide?
Oh, and you just gave away your age, as anyone over the age of 12 knows not to cite Wikipedia as a source.
oneexpert==moron
This comment has been removed by the author.
Intel has had 5W Pentium M CPUs for years now. Even the highest end Core 2 Duo for laptops (2.4Ghz) has a TDP of just 35W. Pure FUD from oneexpert as always.
As was pointed out earlier, these CPUs can be used in desktop PCs if people want to.
Look at these 2400 and 2600 results:
http://www.hothardware.com/articles/ATI_Radeon_HD_2600_and_2400_Performance/?page=4
100% Fragged by Nvidia.
Clovertown vs Opteron:
Fragged in Integer performance:
http://www.techreport.com/reviews/2006q4/xeon-vs-opteron/sandra-mm-int.gif
Fragged in floating point performance:-
http://www.techreport.com/reviews/2006q4/xeon-vs-opteron/sandra-mm-fp.gif
Look at this list of the top five semiconductor companies. Companies like Intel, Samsung and TSMC make the list but of course AMD doesn't. Pathetic.
http://www.theinquirer.net/default.aspx?article=40654
AMD BK Q2'08.
This comment has been removed by the author.
Barcelona is pathetic. Just two Opteron glued together with more cache and 128bit SSE copied from Intel. Pathetic.
They can't even scale that garbage past 2Ghz.Intel has Clovertown 3Ghz shipping today.
Penryn coming this year, already shown running at 3.33Ghz. Penryn featuring revolutionary 45nm high-k process technology. AMD has nothing of the sort.
If AMD ever gets 45nm process working they still don't have high-k technology to reduce the leakage by a factor of 20. Intel showed processors based on this revolutionary process running back in January. AMD hasn't shown anything of the sort yet.
AMD will have to report it's quarterly results soon. Expect another large loss. Intel will report profit in excess of AMD's revenue.
AMD BK Q2'08.
Oneexpert, you need to stop being a fucking moron.
The C2D is a 65 watt piece.
Things are getting SO bad for your fanboi ism retardism that you need to now start including IBM's Power lineup when talking about CPU's, because AMD's are using too much power, have two fewer cores and all around are getting beat by those "P3's glued together".
Imagine that AMD, getting fraged in performance, sales and power utilization (because the AMD chips are only saving 10 watts when at IDLE and when under load consume almost 25 more watts) by Intel's ancient "p3's glued together".
How embarrassing that must be. Its like your sexual performance, not only do you have a small penis, but you have an issue with premature ejaculation as well.
AMD's new add should be
"AMD, the only choice if your computer is doing nothing"
Hey oneretard, why don't you comment about those 250 watt AMD video cards? One AMD component uses more power than an entire Intel computer.
An now AMD is talking about quad crossfire.
1000 watts of power required to produce video.
That's like my entire family's power consumption to drive the video output of a PC.
onemoron,
Care to take a guess how much profit AMD will produce with thoese so called superior products this quarter?
C'mon, what will the global cpu market say those cpu's are worth?
Intel just announced they are a partner in the OLPC programme...
This comment has been removed by the author.
Is there an echo in here?
I keep hearing the same thing come out of onemorons mouth, over and over.
How about it dude, can you answer any of the questions that have been asked of you?
OneExpert said... AMD stands alone in the production of DESKTOP energy saving cpus and platforms.
When saving our planet counts, you can count on AMD.
The BE-2350 and 3800+EE are decent low energy cpus but you lose out on performance. If you try and make a games PC out of one of them then you cancel out any energy savings (especially with a 2900xt).
The minute you start going for faster AMD cpus then power consumption is worse the Core. You only have to look at the quadfx and 6000+ cpus to see how much power AMD's cpu can consume.
Oneexpert, if you had actually read Tom's Article, you would have seen that the top 3 (yes 3 not just one) CPUs for power efficiency were... INTEL chips. In fact there was quite a bit of controversy on Tom's message boards due to the fact that he was so positive about AMD while his own charts put Intel in the lead.
So why is AMD so touted? Well if you look at Tom's methodology, he heavily weighted in the price of the CPU. Now, anyone can see that AMD CPUs are much cheaper.... they have to be or nobody would buy them right now. (To use a Jeach analogy that actually makes sense: If I priced Kia higher than a BMW nobody would buy the Kia).
Also note: Tom took into account the price cuts from Monday for AMD, but he has not taken into account the July 22nd price cuts from Intel.... if he had done the exact same tests on the exact same chips 12 days later his results would be radically different.
So what Tom has proved is:
1. Current Intel CPUs use less energy to get work done than current AMD CPUs.
2. Some AMD CPUs at the extreme low-end (like Semprons) will use less power while doing nothing than Intel CPUs, but as soon as they have to do something the Intel CPUs are much more efficient.
3. A Quad Core Intel QX6700 only uses about 7 watts more power than a dual-core Athlon 6000 at load.... while also being much faster and therefore vastly more energy efficient (right here at bottom of the page
4. AMD CPUs are cheaper than Intel CPUs: Yes, everybody already knows this. While I can't complain about low CPU prices, bear in mind AMD is forced to make its CPUs cheap because they are inferior to what Intel puts out. If AMD's chips actually were vastly superior.... people would pay more for them, they would cost more than Intel CPUs and AMD would actually be making a profit.
5. Tom's is very pro-AMD, so when Penryn benchmarks come out (and I think they will be out before anything about the K10 surfaces), if Tom's is favorable to Penryn, you'll just look like even more of a moron when you call Tom an Intel Pumper.
AMD is forced to make its CPUs cheap because they are inferior to what Intel puts out.
A direct quote of Henri Richard from the Q1 earnings conference call:
"We are pricing our products for what we believe is the value they provide in the marketplace"
See oneexpert? Even the executives at AMD know that they have to lower the prices of their chips because Intel provides a better value.
Or, as we say: You get what you pay for.
Sharikou,
Keep up the good work, its so nice to see all these people stress themselves up to defend Intel, unbelievable:-)when AMD is just doing their thing namely flooding the market with superior CPU's at an absolutely affordable price.
So OneExpert, how does that crow taste???
Rather bitter?
Maybe if you were not a fanboi you wouldn't have to have gobbled that bird down...
OK, this is just freakin' hilarious: From TGDaily
While we often and talk about the good old days of the steamy AMD vs. Intel battle, the truth looms large. But those days of traditional rivalry could be changing. At least Intel itself believes that it is not only AMD the company is competing with: Sources have told us Intel is circulating a document to its employees, one which sums up that current mindset: It's time to move on, Intel has bigger fish to fry.
Apparently the K10 is not striking fear into Intel right now.
Apparently the K10 is not striking fear into Intel right now.
When they can only clock it at 2Ghz that's not suprising. At least in servers, your applications can usually use all four cores.
On the desktop though, a 2Ghz K10 doesn't stand a chance against a 3Ghz K8 in single threaded applications.
K10's 15% IPC increase (AMD's own number) is not enough to make up a 1Ghz clockspeed defecit vs K8 in those single threaded workloads.
Hey Sharikou, what happened to this prediction?
2Q07: Intel server market share drops to 40% as AMD ramps Rev H quadcore. Bulldozer hits hard.
When AMD and Intel report their results for Q2'07 later this month do you think that AMD will have taken 45% of server market share from Intel to leave Intel with just 40%?
I think the only bulldozer that is hitting is the one hitting AMD's fabs that are no longer needed.
bgt
"its so nice to see all these people stress themselves up to defend Intel, unbelievable:-)when AMD is just doing their thing namely flooding the market with superior CPU's at an absolutely affordable price."
So, what do you think how much money will AMD loose in Q2? Scientia says that probably around 400-700M and I agree with him.
Barcelona processors fail to top Intel Core 2
... AMD will not recover processor performance leadership against Intel’s current products,
http://www.vnunet.com/crn/news/2194200/barcelona-processors-fail-top
AMD BK Q2'08.
This comment has been removed by the author.
oneexpert can you FUCKING READ?
Intel's C2D are Desktop processors.
When under load, Intel's C2D processors consume MUCH less power than AMD's processors.
ANyways, with the wee little power savings AMD's CPU's give you under idle, the MASSIVE amounts of power consumed by it's ATI line more than makes up for it.
AMD CPU's: The choice if you wanna do nothing.
AMD GPU's: The power hungry choice!
OneExpert: Moron
You remind me of the 5 year old kid with the fingers in the ears saying "Nu-uh"
OneExpert repeating the following juck :-
Intel only produces power sucking antique designed c2d desktop cpus and platforms.
And who produces the quadfx, 6000+ and 2900xt ? Now that's real power consumption.
The 45 watt desktop AMD cpu stands alone as a energy saving desktop cpu.
The intel c2d is designed to create greenhouse gases that destroy planet earth.
AMD have 2 low efficient parts, how much power do the 6000+, Quadfx and 2900xt use ?
AMDs power saving desktop cpus are cheap too buy, cheap to run, and save power when coupled with a AMD/ATI platform they can save planet earth.
And again, 6000+ Quadfx 2900xt. On one hand AMD have 2 low power slow(ish) cpus and on the other hand heavy power comsuming components that are still not the best performers.
AMDs x2 be 2350 beats out a ex6800 in performance on pdf creation and sells for 1/10 the price while saving tons of power.
6000+ => 86 seconds
be-2350 = 88 seconds
ex6800 = 89 seconds
4000+ = 90 seconds
Looks like pdf creation is not really affected by processor. Lets have a look at other tests :-
CD to MP3 conversion - ex6800 136 secs vs be-2350 247 secs
iTunes - ex6800 103 secs vs be-2350 153 secs
File Compression - ex6800 60 secs vs be-2350 92 secs, and so and so on
As for repeatable, I keep seeing core benchmarks consistently beating AMD cpus.
No amount of intel fanboys can fix the power sucking c2d desktop antiques.
And no amount of AMD fanboys can fix the power sucking quadfx,6000+,2900xt
IBM fastest production cpu 4.7 ghz power6.
Perhaps IBM should make AMD's next cpu, they could use all the help they can get.
Intel best fud, spinola, and marketing departments.
Who remembers, 2900xt release delays, Barcelona is not delayed, Barcelona projected performance graphs, etc etc etc
This comment has been removed by the author.
Oneexpert... do you remember earlier when I told you to make your posts more amusing?
You're failing miserably at it! The stuff you put up is repetitive and annoying.
Now obviously I'm expecting you to (at best) take things out of context and spin for AMD, after all you are a fanboy so honesty is not required. However, if you can't make it amusing, don't bother posting, we've all read your tired old screed about about how one AMD CPU ran 2 seconds faster than an Intel CPU on one benchmark where the hard-drive was the real piece of hardware being benchmarked. Oh by the way... by your own numbers the 6000+ from AMD is barely any faster than parts that are clocked a full Ghz slower. Do you want to go on the record as saying that AMD CPUs cannot scale? Hey Sharikou, Oneexpert is insulting AMD again!
"You cannot find or buy a 300 watt part in the AMD lineup" --> That's like saying you can't find an SUV that only gets 1 mile-per-gallon so all of our SUV's are super-efficient! (Uh oh... that's a Jeach analogy there, but it makes sense). I can find the '4x4' in the AMD lineup, which even according to your main man Tom at Tom's hardware is "We can't fault Vigor for building a system using inferior processors when these are the best an AMD enthusiast can buy.". Let's see.... massively more power than Intel true-quad core, and it's also slower! AMD really is ahead in both areas!
oneexpert said...
When a AMD energy saving cpu x2 be 2350 frags the intel flagship ex6800 on pdf benchmark it does not count(intel fud, spinola).
Way to go, lets ignore the results I posted now that you have your single good benchmark. Let's repeat :-
6000+ = 86 seconds
be-2350 = 88 seconds
The 6000+ is 2% faster at pdf creation than a be-2350 and the 6000+ runs at a 40% greater clockspeed. Now explain to me how that is a useful benchmark because all that tells me is that one of the fastest AMD cpus is only 2% faster than one of AMDs lower speed cpus. I'm impressed.
d 805,d820,d830,d840 ee consume enough power for 4 fxs or 4 6000s.
Lets look at this Tom's Hardware review
quadfx (2 x fx-74) = 261W Idle and 650W total load.
If you look at this review, the D840 uses 232W idle and 344W loaded. It's not the same review but it gives you a fair idea of the sort of power usage you may see.
So far you keep quoting the same benchmark and now you are comparing against Intel's old 90nm CPUs.
intel holds the world record with d840s for consuming the most electrical power in a cpu design and intel is still selling these space heaters.
Refer to above quadfx power figure or any other review on the web that show what a real space heater is like.
My own watt meters and tests confirm you are quoting lies, AMD cpus and platforms consume less power at 100% cpu loads than c2ds.
lol, and you come across as such an impartial poster.
intel has a horrible record of power abusive cpus and platforms.
All I need to say is QuadFx
You cannot find or buy a 300 watt cpu in the AMD lineup.
No you can buy a quadfx with a 650W under load. Imagine if you add multiple 2900xt graphic cards.
I appreciate all the money you intel fanboys spend replying to this blog, I know first hand how much electrical power it takes to fire up your intel cpus and platforms.
I'm using a Core powered laptop to view this blog, much more efficient than my AMD powered games desktop. So need need for appreciation here.
The top 5 supercomputers are IBM and AMD not intel.
And yet the number of Intel systems in the top 500 has increased to 57.8% while AMD has dropped to 21%. Much like their market share has dropped, losing all the gains made recently.
And nice job, looking at just the top 5 systems when its a report about the top 500. Not that I'm surprised that you want to put your own spin on it (like quoting a single benchmark or comparing against old cpus).
Intel expensive to buy and very expensive to run.
AMD are cheap to buy, some are efficient, some are power hogging beasts and after all that they are not the best performing.
AMD cheap to buy....cheap to run....great performance.
Go on quote that pdf creation test again where the fastest AMD cpu is only 2% faster than an AMD running 900Mhz slower.
oneexpert
"VIA fastest speedstep C7, 15,000 times faster than intels."
I'm just curious, how fast is AMD c'n'q compared to Via? Perhaps only 10000 times slower?
AMD's CPU's:
1.) require more processor to run what an intel chip can.
2.) thus the AMD cpu is not under idle.
3.) thus the AMD cpu consumes more power than the Intel chip
Thus once again, oneexpert is simply wrong.
AMD: The right choice if you wanna do nothing.
Alright one last thing for the weekend, I'm going to put on an example of how to properly talk about unconfirmed information:
There are rumors that some of the Penryn NDA's are going to be lifted on July 15 (that's tomorrow as I write this). Since July 15 is a Sunday, I'm thinking they might actually mean 23:59:59 on July 15 meaning the info won't actually be available until Monday.
Note that I said these are rumors. Sharikou & Oneexpert: Also note how I didn't make any massive unsubstantiated claims in this comment since I was actually dealing with facts. Note how I haven't been running around screaming about how Penryn is 10X faster than a tuna sandwich because of the lack of facts (having said that, there is more publicly available info on Penryn's performance than AMD has on K10, but I'll wait for 3rd party evaluations before making judgments). Randy Allen (the AMD guy not the poster here) could sure as hell learn from my example.
So moral of the story is, tomorrow or Monday could be very interesting since we might see a bunch of 3rd party benchmarks on Penryn and get a firmer idea on how fast it really is.
Oh, and one last thing Sharikou-expert, if I'm wrong, I'll actually admit the fact that I'm wrong. Hey Sharikou.... it's July so uh... where the hell is that K10 server you were claiming to buy last month... I'm sure you'll do the right thing and admit you were wrong about that..... or not lol.
All I need to say is QuadFx
Well at least this is a special platform for special people/needs.
The P4/PD and Celeron where/are for everyone. Even one celeron consumed/consume 90W while Semprons did/do 35W.That 3X more power consuming at much lower performance. Sempron where/is faster than Pentium 4.
AMD: The right choice if you wanna do nothing.
Well in my company server get maxed out 20% of 24/7 time.
The users are using their Pcs with basic applications 3 hours out of 8 hours.
Then AMD must be the right choice. Upsss, wait we have just P4 without any power consuming trick. Maybe we could trade?
AMD is using power saving technologies in desktop/servers since 2003, Intel 2006/2007.
Full load:
PD805 474W at 4.1Ghz
PD805 258W at 2.66Ghz
Doing nothing:
PD805 259W at 4.1Ghz
PD805 171W at 2.66Ghz
More 90W and both doing the same, nothing.
X2 6000+ 11.79W
X2 3800+ 6.14W
More 6W and both doing the same, nothing.
In the end maybe AMD is trying to follow the Intel (past) footsteps, what’s wrong, if that didn’t stop people buying Intel CPUs, why will now they stop buying AMD?
In my country laptop with AMD cpu and 2GB ram sell more than laptop with Intel cpu with 1GB ram, amd system cost the same or is even cheaper. Nobody cares the CPU brand they care more the installed RAM or the HDD size.
Maybe that’s why Intel is afraid of Samsung… and nvidia … and any others that can do better than they. Who knows some super chipset from nvidia could ruin Intel dominant chipset businesses like Via already did in the past.
Aguia said :-
"Well at least this is a special platform for special people/needs.
The P4/PD and Celeron where/are for everyone. Even one celeron consumed/consume 90W while Semprons did/do 35W.That 3X more power consuming at much lower performance. Sempron where/is faster than Pentium 4."
Then let's compare quadfx against it's equivalent. How much more power does quadfx used compared to an Intel Quad and still not out perform it ?
The fact that it is a 'special' platform does not alter the fact that it has set a record for power abusive cpu / platform outdoing anything Intel has managed in the past.
Aguia said :-
"AMD is using power saving technologies in desktop/servers since 2003, Intel 2006/2007.
Full load:
PD805 474W at 4.1Ghz
PD805 258W at 2.66Ghz
Doing nothing:
PD805 259W at 4.1Ghz
PD805 171W at 2.66Ghz
More 90W and both doing the same, nothing.
X2 6000+ 11.79W
X2 3800+ 6.14W
More 6W and both doing the same, nothing."
Only a couple of points.
The PD805 at 4.1Ghz was 50% overclocked compared to it's standard speed, I'd be interested to see the power figures on an AMD cpu with a 50% overclock. If I remember correctly the D805 was also Intel's old 90nm technology.
The X2 6000+ 11.79W & X2 3800+ 6.14W idle values appear to be from this review where the power was measured at the cpu socket. Here's some other values :-
CPU --------- Idle -- Load
6000+ ------ 11.79 - 119.47
3800+ EE ---- 6.14 -- 46.45
6300 (l2) --- 8.62 -- 37.36
6300 (b2) -- 12.02 -- 45.49
x6800 ------ 20.64 -- 66.97
QX6700 ----- 24.28 - 129.66
The new L2 stepping is a good challenge for AMD's EE cpu. Looks that that power saving technology that intel started using in 2006/2006 is coming along very nicely. (nice load on the 6000+ it's almost catching up with a quad )
This comment has been removed by the author.
oneexpert needs to be a member of AMDZONE.com, The ultimate sanction of AMD fannies. The moderator will welcome you with arms wide open.
I love it when I read over there that, hey It might not be faster but it would be more elegant.
THe K10 will do the work more elegantly, slower but elegantly....
WTF IS WRONG WITH THESE FANNIES? Elegantly? lol is that all they can resort to now?
I laugh my ass off everytime i read something like that.
oneexpert said :-
"Chuckula the data you presented proves my point perfectly;
2-fx74 are 250 watts
2-d840s are 688 watts"
Nice attempt at twisting what others have posted :-
You are comparing 2 of AMD's latest cpus (fx74) running at idle with 2 of Intel's old 90nm cpus running at load. Your figure should read 650W total for the pair of fx74.
oneexpert said :-
"2-qx6700 are 864 watts
2-fx74s at 650 watts"
so 4 AMD cores use 650 watts vs 8 Intel cores using 864 watts. It should be 2 fx74 using 650W vs 1 qx6700 using 432W.
So first you compare against old Intel cpus and then you double the core count to show higher power usage. Nice & impartial... not
oneexpert said :-
"As you can see from chuckulas data intel cpus suck nearly twice the power of AMD cpus.
1-AMD x2 be 2350=45 watts
AMD modern technology with space age results.
AMD stands alone with energy saving desktop cpus."
As I can see from your data you are happy to post any junk to support your arguments about Intel using twice the power.
I also find its funny that the new be-2350 (2.1Ghz) & be-2300 (1.9Ghz) are rated 45W while the older 3800+ EE (2 Ghz) was rated 35W but I suppose that is progress for you.
aguia
"Full load:
PD805 474W at 4.1Ghz
PD805 258W at 2.66Ghz"
Why are you comparing full system power usage to only CPU usage for AMD?
oneexpert
"Chuckula the data you presented proves my point perfectly;
2-fx74 are 250 watts
2-d840s are 688 watts"
For an "expert" you are kind of dumb that you didn't make the difference between full system load and CPU only load.
Why are you comparing full system power usage to only CPU usage for AMD?
Sorry didn’t noted. It was already late...
And don’t forget the only CPU usage on AMD includes North Bridge and IMC. Intel doesn’t. From Toms:
Therefore, AMD's processors retain the crown for lowest power consumption when idle, aided by the fact that chipsets for the AMD platform also tend to have lower power requirements.
That’s why toms declared AMD the winner. Even if Intel came out on top, it wouldn’t if the all system was being evaluated.
And about the L2 stepping, isn’t the B2 6300 one 4MB processor with half cache disabled, and the L2 stepping one native 2MB processor, and so the power consuming difference.
I think B2 6300 and B2 6400 2MB consume the same of one 6600 4MB because of that, and that’s also why there is no L2 stepping with 4MB cache?
Some time ago it was popular to post real-world users experience with their CPUs. Here is one: e6300 based home server takes on average 77 Watts.
Here is another Hoho,
Green Machine
The e6400 98.6W
The 3800+ 67W
The 4600+ 78.9W
Aguia said :-
"Therefore, AMD's processors retain the crown for lowest power consumption when idle, aided by the fact that chipsets for the AMD platform also tend to have lower power requirements.
That’s why toms declared AMD the winner. Even if Intel came out on top, it wouldn’t if the all system was being evaluated."
From the same page of Tom's review.
"Under full load, the picture changes, and we can see that Intel has been able to improve its processors' efficiency over the last few years. AMD, on the other hand, has lost ground in this discipline, due to the fact that it is still using the same architecture."
As a number of people have been saying - AMD For Efficiency, Intel For Performance
Nehalem is coming in H2'08: http://www.vr-zone.com/?i=5113
AMD is finished. K10 is fragged by Clovertown. Still, Penryn is coming this year and Nehalem is coming next year.
Intel is very good under full load (only cpu usage) like hoho said, not the complete system.
For the complete system AMD wins, that’s why Toms in their review:
Intel Disappoints in Energy Efficiency
Power Dissipation: AMD Still in the Lead
Energy Index: AMD Unbeatable
Drawing a Balance: Core 2 with L2 Stepping Disappoints
Efficiency: Intel's Core 2 Unbeatable
Overall Results: L2 Stepping Takes the Lead
Conclusion: AMD still Offers Lowest Power Consumption despite L2 Stepping
Nehalem is coming in H2'08
They have to. So where would be the tick-tack-tock, Intel was talking about.
Conroe July 2006. Penryn July 2007. Nehahem July 2008.
:)
This comment has been removed by the author.
Every time you post oneexpert you look more and more like a fucking retard.
Get your helmet back on, get back on the short bus and shut the fuck up.
Its been proven time and time again you have no clue what you are talking about.
Buy Amd and you'll get better power efficiency when idle (unless you get a quadfx). Buy Intel and you get performance, apparently very hard to understand.
Well we are talking about oneexpert here... not quite the fountain of intelligence and understanding one would expect for a person with reading skills.
And all those vapor ware and paper launches of nehalem and penryn simply will not change intels position.
This is hilarious. Intel's execution since 2006 has been near perfect. The Core Micro-Architecture was early (Woodcrest in June, Then Conroe/Merom in July). Quad cores parts were pulled in by three months.
This is in stark contrast to AMD with the constant R600 delays and K10 arriving later than the original "Mid-2007" and only at 2Ghz.
45nm high-k Penryn is set for introduction in H2'07 and they will deliver Penryn in that timeframe.
Post a Comment
<< Home