Intelers are in denial
As I have shown in this comparison: the C2D and K8 architecture are neck to neck in terms of performance. Except the SSE advantage, the only trick C2D has is cache. Once the cache is reduced to the same level of K8, the C2D loses its advantage. Similarly, when C2D's cache is not enough, it's also slower than K8. As we noted earlier, 3GHZ Opteron conclusively frags 3GHZ Woodcrest in both integer and floating point computation.
Essentially, Core 2 and K8 cores have equal IPC in terms of integer, and Core 2 is at least 15% slower on FP.
K10 core will have 40% IPC lead over Core 2 on integer (per core).
K10 will have 200% IPC lead over Core 2 on floating point (per core).
posted by Sharikou, Ph. D. @ 4/15/2007 02:01:00 PM
121 Comments:
"Ph"ony "D"octorate
You haven't shown me a thing. Show me a the still unseen AMD product running in a machine side by side with a Clovertown.
Why are there no screen captures no FPS, there simply isn't ANYTHING.
Like your PhD its just totally Phony
Sharikou
Essentially, Core 2 and K8 cores have equal IPC in terms of integer, and Core 2 is at least 15% slower on FP.
To begin, it is understood that AMD's platform has an advantage due to HTT in 2P+ systems, but that is not what Sharikou's post is about.
Sharikou's post is about per core performance, and as such he should be trying to show a per core comparison (as close as possible with dual cores).
So lets look at 1P performance, eliminating the HTT advantage over Intels FSB, while showing a core to core comparison.
Looking on SPEC for similar systems I found...
1. AMD FX62 CFP2006 Rates
2. Intel E6700 CFP2006 Rates
3. AMD FX62 CINT2006 Rates
4. Intel E6700 CINT2006 Rates
I would like to note that the AMD system is using DDR2 1066 CL5-5-5, and Intel is using DDR2 800 CL5.
AMD-FP 18.8 in SPECfp_rate_base2006
Intel-FP 25.1 in SPECfp_rate_base2006
AMD-INT 21.8 in SPECint_rate_base2006
Intel-INT 29.1 in SPECint_rate_base2006
These numbers tell a different story than Sharikou's post of how AMD and Intel compare on a core to core comparison.
It is looking like Intel has about a 33% advantage in Floating Point performance, and about a 33% advantage in Integer performance.
Again it is understood that AMD has an advantage in 2P+ systems due to HTT, but these resuts show a different picture than our host is trying to present when comparing the microarchitectures of Conroe and K8.
Notes
1. I am hoping someone could explain the effect of the higher latency on the AMD system.
2. If there is an error please point it out.
Thanks.
enumae said..
"I would like to note that the AMD system is using DDR2 1066 CL5-5-5, and Intel is using DDR2 800 CL5.
"These numbers tell a different story than Sharikou's post of how AMD and Intel compare on a core to core comparison".
Don’t misunderstand! What sharikou trying to explain is “when the L2 cache memory of the C2D is reduced to equivalent with the K8 L2 Caches” the C2D loses its advantage. Similarly, when C2D's cache is not enough, it's also slower than K8.
Note:
L2 cache memory is totally different with the DDR2
What the hell are you smoking Shakirou?
I found AMD's top secret K10 benchmarks!
http://img476.imageshack.us/img476/360/k10eq5.jpg
Here's some good benchmarks comparing Core 2 Duo performance at 3Ghz with 1mb, 2mb and 4mb of L2 Cache. All 3Ghz w/ 1333mhz FSB. There's an Athlon X2 3600+ at 3Ghz as well for good measure.
http://www.pconline.com.cn/diy/cpu/reviews/0704/993288.html
Then the question remains: Why didn't AMD pump the cache to a level where to take the lead? It is one of three:
1- More cache means more expensive CPU. But then even if this is the reason, they can produce a processor and call it extreme, turbo, or you name it. Why do not they do it?
2- It could be That cache does not make a a noticeable difference in AMD performance. This is a drawback if true. If this is the case, no need to bother the answer as why they don't do it.
3- Both points above.
sharikou .. I am an amd fan. I am eager to upgrade to next AMD cpu killer. But I still need to be convinced there will be such one. Can you tell me if the cache is the trick, why don't they just apply it?
"Can you tell me if the cache is the trick, why don't they just apply it?"
AMD doesn't have as much capacity as Intel, which can afford lower yield and more transistors while still remain profitable. In the semiconductor industry, size matters.
In other words, Intel's large cache and processing technologies could have resulted in much better system performance if they were applied to AMD's K8/K10 design.
Fifty percent faster than Kentsfield
"K10 part called Barcelona is up and running. It is not ready for launch but AMD already have the first part ready.
We don't have many numbers but at least in specfp_rate2000 Barcelona ends up a bit less than 50 percent faster than Intel quad core codenamed Kentsfield.
We are quite sure that this is the best score but it definitely sounds impressive. AMD chaps are very confident that K10 marchitecture and native Quad core is the way to go and Intel will probably have to pay the price for duct taping its chips. AMD will come to collect in second half of 2007."
http://www.fudzilla.com/index.php?option=com_content&task=view&id=543&Itemid=1
"Al-Saqer said..
Then the question remains: Why didn't AMD pump the cache to a level where to take the lead?"
This is the answer to your question.. check it out.
http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543_544~115794,00.html
"Quad-Core AMD Opteron processors are packed with core and cache enhancements designed to improve performance on a range of server and workstation applications. Cache-sensitive transactional applications such as Web, database and email servers can benefit from the addition of a 2MB shared L3 cache. Simulations conducted in AMD laboratories indicate that certain database applications will see performance improvements up to 70 percent and certain floating point applications will experience performance gains of up to 40 percent over platforms powered by current dual-core AMD Opteron processors."
randy allen,
That was quite funny.
It’s funny but is true.
If you joint the best intel cpu + best intel chipset + best intel gpu
VS
best amd CPU + best amd chipset + best amd/ati gpu
In one game that would certainly do the 500% difference in performance or more.
Amazing isn’t it?
how bout this one
QX6800 extreme + Intel® 965 Express Chipset + Intel Extreme Graphic
VS
Amd 3000+ (lowest amd processor) + AMD 690 chipset + R600 series graphic card
who you think the winner? very funny.. ;-)
pezel: This is the answer to your question.. check it out.
http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543_544~115794,00.html
This answers what effect do the cache have in performance. If I understand the article, they will increase cache in quad cores, making them perform better in multi-threaded applications. But why do not they have dual and single core processors with higher cache as of yet to respond to the C2D?
Dear Ph(oney)D,
Do you REALLY think Intel is going to be where they are today when AMD finally releases the K10?
Its wonderful you can compare something that isn't even out yet, with something that is.
Mayhap you forget that Intel is contining its release of updated CPU's over the next year or so.
Hell by late last year, the K10 is still having serious thermal issues which are limiting its clock speed quite a bit...
But the phony PhD man keeps claiming Intel will be bankrupt by Q2 2008.
Al-Saqer said..
But why do not they have dual and single core processors with higher cache as of yet to respond to the C2D?
opteron X2 64bit is the amd respond to Pentium 4 & Pentium D. Meanwhile, AMD Barcelona is the amd respond to C2D&C2Quad. Just wait and see..
pezal said:
opteron X2 64bit is the amd respond to Pentium 4 & Pentium D. Meanwhile, AMD Barcelona is the amd respond to C2D&C2Quad. Just wait and see..
And right now they are poor responses. Everyone keeps talking how AMD is going to come out with CPU's that are going to "crush" and "bk" Intel, but fail to mention the fact that Intel isn't staying still thru the process either. Lets not forget Intel has already mastered the 45um process, and has stuff like Penryn, Wolfdale and Yorkfield quickly approaching, not to even mention Nehalem, Westmere and Gesher, which are going to move quickly to the 32um platform (as well as contain integrated memory controllers).
Sure AMD is moving forward, but so is Intel.
sharikou said: "Once the cache is reduced to the same level of K8, the C2D loses its advantage"
Good point! Intel's roadmap is suggesting that future designs will have less cache. With the trickery, AMD can finally run Intel to the ground!
Here's the link:
htt://www.CPU.com/Intel-roadmap.xml?8892absjj/I-am-an-idiot-if-I-actually-follow-this-link=78993-abse?e.html
Read & weep intelers!!!!
http://articles.techrepublic.com.com/2100-1035_11-6176204.html?tag=nl.e019
"Only AMD said...
Read & weep intelers!!!!
http://articles.techrepublic.com.com/2100-1035_11-6176204.html?tag=nl.e019"
What relevenance does that article have at all to Sharidouche and his lemmings delusions? All it talks about is Intel moving forward with it's plans in China, the lack of R&D due to engineering skills *IN CHINA*. The linked article says China's education system and it's engineers are no where near as skilled as those coming out of the US.
So please show me what that article has to do with K10 .vs Core2 performance?
onlyAMD & Tim
This is why Intel is not going to do research and development on processors in China.
Intel’s proprietary business model will lose ground this year in many facets of modern computing from mobile to enterprise in the next cycle of offerings. The CPU is just one cog in the wheel. Real world workloads will depend on more than just calcs per clock. MB’s, compilers, graphics and chipsets optimized to enhanced algorithms will mature with 64-bit in 2007. There is enough arrogance and pride at Intel to more than make up the difference in short order. With billions of $ on the line; expect hyperbole from both camps. Enjoy the ride.
evil said..
"Lets not forget Intel has already mastered the 45um process, and has stuff like Penryn, Wolfdale and Yorkfield quickly approaching, not to even mention Nehalem, Westmere and Gesher, which are going to move quickly to the 32um platform (as well as contain integrated memory controllers"
But once AMD response Intel will be in trouble
I think, 2xBarcelona sc1207 (4+4=8 Cores) is more than enough to beat intel 65nm/45nm forever.
8 Cores vs 4 cores?
"enumae said...
onlyAMD & Tim
This is why Intel is not going to do research and development on processors in China."
Yeah I read that one after I posted - but it really has nothing to do with Shariboob's claims....
At IDF Beijing Intel unveiled a little more about Penryn performance; it compared a quad-core 3.33GHz (1333MHz FSB) Yorkfield with 12MB of L2 cache (2 x 6MB per dual core die) to a quad-core Core 2 Extreme QX6800 2.93GHz (1066MHz FSB) Kentsfield with 8MB of L2 cache (2 x 4MB). According to Intel’s own benchmarks, Intel saw a 15% increase in imaging related applications, 25% in 3D rendering tests, greater than 40% in games, and a greater than 40% increase in video encoding performance when SSE4 support was utilized.
It was fun while it lasted. RIP AMD.
"All it talks about is Intel moving forward with it's plans in China, the lack of R&D due to engineering skills *IN CHINA*. The linked article says China's education system and it's engineers are no where near as skilled as those coming out of the US."
Oh, yes, Intel is such a humanitarian company, that it would rather hire engineers from China to work in U.S. than to have them work in China with lower wages and benefits.
Wow. Thanks for those Penryn numbers. Penryn is going to be an awesome upgrade from Core 2 on the desktop. Still socket 775 too. What Intel hasn't mentioned is how high Penryn can clock. But thanks to it's revolutionary high-k 45nm process Intel can clock these massively high will maintaining a low TDP. Not like AMD's crap. Intel does not mention shipping clockspeeds because they will adjust them as needed to totally frag Barcelona.
You think things will look ugly for AMD when it reports it's financial results later this week? It's only going to get worse for AMD. There's no way AMD can make it out of 2007 alive. If they want to, they will have to sell one of their fabs to raise captial.
AMD 6000+ dual core using more power than a quad core Intel Extreme Edition. Pathetic. AMD's 4x4 platform is a great deal, heat your whole house this winter with your PC!
I think, 2xBarcelona sc1207 (4+4=8 Cores) is more than enough to beat intel 65nm/45nm forever.
8 Cores vs 4 cores?
You seem to forget that Nehalem will scale past 8 cores and is coming next year. Intel can committed itself to an incredibly aggressive roadmap with the sole purpose of wiping out AMD.
Pezal said:
I think, 2xBarcelona sc1207 (4+4=8 Cores) is more than enough to beat intel 65nm/45nm forever.
8 Cores vs 4 cores?
Maybe you haven't seen the fact that Intel is already showing the early revisions of the Peryen performance, the fact that Intel too is planning 8 core and that Intel has a LOT more money to throw into R&D than AMD does...
I hope AMD has a response, much like the Athlon was, because all it does is make Intel work harder better and faster. Keep pushing the gorilla, and eventually he pushes back. Well he is pushing back...
"You seem to forget that Nehalem will scale past 8 cores and is coming next year. Intel can committed itself to an incredibly aggressive roadmap with the sole purpose of wiping out AMD".
Im sure, it is also will be remaining the same because during the time AMD also got a new processor called Shanghai. Shanghai will be manufactured in 45 nanometer and will hit the higher speeds. Despite more L3 cache memory the new CPU will have some minor architectural changes. It is a native quad core but you will be able to put two of them together for eight cores.
So,
2xShanghai (4+4=8cores) vs Nehalem (4 cores)? Is this what you meant?
http://www.itnews.com.au/newsstory.aspx?CIaNID=48810
Nehalem will introduce two processing threads for each core, up from the current single threaded cores. Intel said it is currently working on 8-core processors, but might introduce larger chips in the future.
8 core, 16 thread Nehalem ready to frag AMD all over. AMD BK Q2'08.
"Mimicking the way that AMD processors are designed, Nehalem will embed the memory controller onto the chip. It is currently part of the front side bus. In another move that follows AMD's lead, Nehalem will deliver an integrated graphics processor"
Looks like intel is stealing amd architecture.. However, just wait and see what will going to happen in 2008 ;-) gtg
With Yorkfield @ 3.33GHz offering more than 40% more performance than a QX6800 @ 2.93GHz, Barcelona doesn't stand a chance. Not to mention, it's puny 2.3GHz clock speed won't be much of a help.
While AMD is sucking red, Intel is making billions in profit. When AMD's stock prices hit below $5, Nvidia and Intel will be picking up the scraps. Hector Ruiz better prepare his new resume LOL.
pezal said...
Looks like intel is stealing amd architecture.. However, just wait and see what will going to happen in 2008 ;-) gtg
AMD is tiny compared to Intel. AMD only exists because they consistantly innovate and force Intel to play catch up, over and over again.
pezal, need it be said the ONLY reason AMD exists is because they copied Intel from the get go?
While AMD is sucking red, Intel is making billions in profit.
Whilst at the same time they are pouring billions into R&D to bring us cutting edge CPUs. It's just not going to stop. Penryn this year, Nehalem in 2008, Westmere in 2009 and Gesher in 2010!
All this while keeping CPU prices affordable. Prior to Core 2 Duo AMD was charging $300 for Athlon 64 X2 3800+, just milking customers for profit. This is deceitful and despicable.
AMD BK Q2'08.
At IDF Beijing Intel unveiled a little more about Penryn performance; it compared a quad-core 3.33GHz (1333MHz FSB) Yorkfield with 12MB of L2 cache (2 x 6MB per dual core die) to a quad-core Core 2 Extreme QX6800 2.93GHz (1066MHz FSB) Kentsfield with 8MB of L2 cache (2 x 4MB). According to Intel’s own benchmarks, Intel saw a 15% increase in imaging related applications, 25% in 3D rendering tests, greater than 40% in games, and a greater than 40% increase in video encoding performance when SSE4 support was utilized.
"It was fun while it lasted. RIP AMD."
The 40% increase was WITH SSE4. Yorktown does not have SSE4. that 40% is bullshit.
Boner said: ...it compared a quad-core 3.33GHz (1333MHz FSB) Yorkfield with 12MB of L2 cache (2 x 6MB per dual core die)...
So it is true what Sharikou said then? The only difference is CACHE! 12MB OF CACHE! Good Lord!
So if AMD did the same, they should see a 432432034% performance boost, no?
I wish AMD follows Intels path, and just keeps adding all sorts of cache... then the Barcelona should always outperform the Intels.
That's a very easy route to follow... adding cache on the die.
How-about-Intel showing us their new architecture and not the Pentium III Xeon Balony Intel is trying to push on us...
(See picture at: http://www.theinq.com/default.aspx?article=38961 LMAO!! Sharikou was correct again!)
And what is this? Intel's latest chipsets don't come with DirectX10? So who would want them?
http://www.theinq.com/default.aspx?article=38963
Tsk Tsk... Tick Tock... straight into the grave...
The 40% increase was WITH SSE4. Yorktown does not have SSE4. that 40% is bullshit.
Of course it does. Penryn has had SSE4 the whole time.
And what is this? Intel's latest chipsets don't come with DirectX10? So who would want them?
Better than AMD's awful chipsets that don't even fully support Direct X 9.0C with shader mode 3. All the latest games that need shader mode 3 will not run on AMD's antiquated chipsets. The only company with Direct X 10 hardware is Nvidia.
Clearly, Nvidia is the leader in graphics, Intel the leader in CPUs. Intel and Nvidia are clearly the smarter choice!
How-about-Intel showing us their new architecture and not the Pentium III Xeon Balony Intel is trying to push on us...
Unlike AMD? K10 is still based on the ancient K6 architecture. AMD based K7 and K8 both off the K6 architecture. AMD didn't even design the K6 architecture themselves, they bought it from another company.
Tough times for AMDers. Core 2 has a commanding lead over AMD, and Penryn will extend that lead by a further 40%.
Intel won't even need Nehalem to finish AMD.
AMD BK Q2'08.
Randy Allen said...
...
Better than AMD's awful chipsets that don't even fully support Direct X 9.0C with shader mode 3. All the latest games that need shader mode 3 will not run on AMD's antiquated chipsets. The only company with Direct X 10 hardware is Nvidia.
Clearly, Nvidia is the leader in graphics, Intel the leader in CPUs. Intel and Nvidia are clearly the smarter choice!
...
Unlike AMD? K10 is still based on the ancient K6 architecture. AMD based K7 and K8 both off the K6 architecture. AMD didn't even design the K6 architecture themselves, they bought it from another company.
Tough times for AMDers. Core 2 has a commanding lead over AMD, and Penryn will extend that lead by a further 40%.
Intel won't even need Nehalem to finish AMD.
AMD BK Q2'08.
Hahahaahaha ..., i'm seeing Sharikou-ish all over the place of your comments. Nice one! :)
Well, Intel just posted earnings greater that AMD's entire revenue.
When's that BK again?
I think he believes if he says it enough it will come true...my 16 year old has a better grasp on reality than Sharidouche....
In case you haven't noted the recent trend, he doesn't post intel earnings announcements anymore because they are positive. He tries to give us the scoop from inside AMD -- the only thing that will keep the people there motivated is to keep telling themselves that they are the best. I bet that AMD regulates that everyone's home page must be this blog, as it is the sole site that has only positive news about AMD.
Interesting way to look at Intel's earnings:
If Intel spent their quarterly net income just this one time, they would be able to purchase 21.7% of AMD-ATI. But hey, they need that money to plow back into R&D and capital equipment upgrades.
So let's look at disposable income. If they spent the $400M they spent last quarter on INTC buyback on AMD at current prices, they could buy 5.4% of AMD-ATI. So for one quarter of skipping buyback, Intel could purchase enough AMD to influence them, not hurt their pocketbook, and not get in trouble with SEC, FTC, or anyone else, because it is still a small minority share...
Perspective?
Bubba said...
When's that BK again?
2Q08. Please try to keep up.
Scott said...
In case you haven't noted the recent trend, he doesn't post intel earnings announcements anymore because they are positive.
Says who? Intel? The same Intel that has been caught blatantly cheating in their benchmarks over and over again?
IDF: Penryn 40% faster than C2D extreme! - DEMOED!!
http://www.macworld.com/news/2007/04/17/penryn/index.php
Looks like Barcelona is DOA. Assuming ofcourse that it actually arrives.
Meanwhile Intel Q1'2007 reports solid quarter. It must be a BK trend indeed to post solid quarters and then a year later without making any sense, file chapter11.
I guess its time for AMD to lower some more prices to get rid of invetory and allow Sharikou to have at least something he can easily spin as good news.
Can't wait for bloody Thursday.
Interesting fact:
If Intel shut down it's fabs today, and kept all spending at current levels - no layoffs, reductions in captital spending (like AMD) etc. - it could continue to operate for nearly five quarters without drawing one cent from it's credit line.
Do the math penix. The only company heading for BK is AMD.
Says who? Intel? The same Intel that has been caught blatantly cheating in their benchmarks over and over again?
What? You mean like AMD too? Go to the AMD website, under CPUs and look at the Turion X2 and Athlon X2 benchmarks. They include lots of graphical tests and use different graphics solutions.
In servers, they post a few select benchmarks where Opteron beats Woodcrest. Of course, in these benchmarks, they fail to include a quad core Clovertown. Except for one benchmark. They include a Clovertown in one benchmark and that's only because the test makes heavy use of Opteron's IMC, easily allowing them to win.
Saying AMD is not guilty of this is plan rubbish!
Either way, did you see those Intel results? Not bad for a company that will be BK in Q2'08! 1.6bn profit.. that's more revenue than AMD will report this quarter!
LOL Why don't AMD add more cache to speed up their Barcelona ask Kingrichard?
WHY not little richard, because AMD is on 90nm and transitioning just now to 65n. While INTEL has
been on 65nm for almost 2 years. INTEL holds a 2x density advantage and thus for the same silicon
area can cram 2x the cache. Add on the fact INTEL as 4 65nm factories and 3 45nm factories in
construction while AMD has just ONE factory means a capability of 8x factor difference in transistor
packing density. ANy questions why AMD loses money and INTEL makes money while also still holding a
3x in unit volume and 2x in cache? Its because of their techology advantage and manufacturing
advantage.
How about them numbers there! $1,610,000,000 profit on 8,900,000,000 sales. That while completing
ramp to 65nn CPUs, finishing R&D on 45nm, building 3 additional 45nm factories, launching multiple
45nm CPU designs
Oh, the competition dropped prices 50%, intel followed and still made 1.6 BILLION profits. Lets
laugh at AMD's results later.
Did you see the gross margin forecast.. it went UP! Damm that sure looks like a company in trouble
and on the way to BK..
Barcelona is too little to late. To bad it didn't come last year and too bad AMD didn't invest 5
billion manufacturing 2 years ago too to make them a realy worthy competitor in x86. Lets look at
how grimm the disadvantage is
Barcelona on 65nm is ~ 283mm^2 with a palty 4 MB of L2/L3 cache
Penrym on 45nm is ~ 107mm^2 with honking 6 MB, multiply by 2x to get a quadcore you get 214mm^2 and
12 MB of cache
Factor in INTELs 4:1 ratio of 65nm factories and 4:1 rato of 45nm factory and figure INTEL can pump
out between 4 to 10x more CPUs then AMD.
"Ph"ony "D"octorate what is your explanation for these results and how will that quad-core Barcelona
change things around. You see them Penrym performance numbers from IDF?
Dude AMD is sooooooooooo in trouble!
Tick (Yonah ), Tock ( Merom ), Tick ( Penrym ), Tock ( Nehalem ), goodbye AMD!
Tick! Tock! Tick! Tock! Boom!
That's the sound of AMD going BK while Intel can still earns 1.6bn in profit!
I like this quote:
"Congratulations to an excellent quarter... you've manage to reduce your competition into rubble". - Joe Osha, Merrill Lynch's semiconductor analyst at Intel's Q1'2007 earnings report.
AMD BK Q2'08.
Oh, wait... the string of AMD bad news isn't over.
Now iSupply is saying that AMD may have lost 1/3 of its market share! we're now back to pre-2003 levels.
http://www.tgdaily.com/content/view/31666/118/
So much for breaking the monopoly.
I can't wait for Sharikou's spin on this one. The harder it is the funnier it gets.
Hey Azmount Aryl, what happend to AMD gaining further market share?
I previously estimated Intel would have an 80% market share by the end of '07. Intel it certainly performing better than I thought; it's market share is already at 79.5% according to iSuppli. Perhaps even 85% is achievable by Intel this year.
It's now clear that AMD is in dire trouble.
I can't wait for Sharikou's spin on this one. The harder it is the funnier it gets.
$10 says it'll be story on how Intel is pulling an Enron!
I bet it's his classic "recoiling to make a huge comeback".. or "pulling back to make a punch".. that was funny. I never heard of any company utilising such clever strategy. oh, wait. there's Transmeta.
Allright time to set you guys straight. Penryn, oh Penryn. It will have some architectural tweaks to improve IPC, add SSE4 (not SSE4a like AMD's new core). It will also work on SKULLTRAIL! Yes, Intel is readying Skulltrail (like AMD's 4x4). So if you guys really want to see what Intel is cooking up google "skulltrail", "Yorkfield" and "Penryn".
http://www.informationweek.com/news/showArticle.jhtml?articleID=199001370&pgno=2&queryText=
In other words, AMD is in for a rough ride.. and I'm no Intel fan either (not to say I'm an AMD fan as well).
"The first quarter earnings results are likely to reveal just how brutal the current market environment in the microprocessor industry really is. While Intel said earlier today that it the company was able to keep declines in average processor selling prices to a minimum, AMD warned the financial community last week that the firm saw “lower overall average selling prices and significantly lower unit sales, especially in the resale channel” during Q1.
Prices for AMD’s desktop processors fell by an average of about 50%, according to a recent TG Daily analysis . iSuppli told us today that these price cuts were not enough to maintain market shares for AMD. Based on preliminary findings, iSuppli vice president Dale Ford estimates that AMD’s overall market share in the microprocessor industry may have declined 4.7 percentage points from 15.7% in Q4 2006 to 11.0% in Q1 2007. The analyst believes that Intel was able to regain market shares in the same time frame: “On a preliminary basis, I would estimate that Intel increased its market share in microprocessors from 75.7% in Q4 to 79.5% in Q1,” Ford told TG Daily. He described the exchange of market shares as “a big swing” and said that the market share shift was bigger than the firm initially had estimated."
So much for Intel going BK, more like AMD going BK.
Penryn is just a Kentsfield shrink with 1) larger cache, 2) SSE4, 3) deeper power saving. Other than SSE4, the performance boost seems to be only 10~15% for 20% higher clockrate and 30% faster FSB speed. Such scaling is terrible. Not to mention the faster FSB will require a more power hungry north bridge, totally negating the better power saving of the CPU.
In other words, performance wise, in most cases Penryn @3.3GHz will still lose to K10 @2.6GHz, and take more power due to its need of high power north bridge.
Of course there will be cases where Penryn outperforms K10; for example, when exotic SSE4-specific instructions are used, or when cache trashing happens at above 2MB (extremely rare). I'm sure Intel will persuade/pressure all benchmarks to favor Penryn, much like what it did for Core 2.
Market-wise, Intel has been doing better than AMD since there's PC, not because it had better products, but because it had more stupid customers. However, every major release since AMD's K6 has been giving Intel heavier and heavier punches. My guess is people are getting smarter more often.
abinstein
"Penryn is just a Kentsfield shrink with 1) larger cache, 2) SSE4, 3) deeper power saving"
You forgot about twice as fast SIMD shuffling and improved FPU that is around 30-50% faster in most floating point calculations and 2x faster in taking square root.
"Other than SSE4, the performance boost seems to be only 10~15% for 20% higher clockrate and 30% faster FSB speed"
Where are your math skills? It has less than 15% higher clock, not 20, FSB also less than 30% faster.
"Not to mention the faster FSB will require a more power hungry north bridge"
What is your source of information?
"In other words, performance wise, in most cases Penryn @3.3GHz will still lose to K10 @2.6GHz, and take more power due to its need of high power north bridge"
Same as above. Also I'm sure that in single socket 2.5GHz k10 will not be able to compete with 3GHz Pentyn in vast majority of tasks. It simply doesn't have so much better IPC. In fact I'd be surprised to see K10 beating Penryn at most tasks at the same clock speed. Sure, it will beat it in some things but most definitely not in majority.
The quoted 40% more performance goes for the FSB limited benchmarks, not real world applications. When AMD compared 2P Clovertown with two 2.66GHz quadcores on 1066Mhz FSB against 2P K10 with up to three times as much bandwidth and they saw only 40% improvement I'd say things are not exactly good for them.
"I'm sure Intel will persuade/pressure all benchmarks to favor Penryn, much like what it did for Core 2"
What pressure? People used the same benchmarks they always had.
Also, you might want to reread the book you were quoting before about the ILP, I wasn't talking about imaginary perfect CPU that was discussed there.
abinstein, go back to 5th grade math. God, are all AMD fanboys dumb as you?
"Not to mention the faster FSB will require a more power hungry north bridge"
What a load of bull.
Sharikou is becoming a fairly sorry looking individual, rather quickly.
He created this blog at the height of AMD power when they scored a big hit against Intel, and at the time the future for AMD couldn't have been brighter. Even the title "64 bit computing" was at the time a hidden jab at the makers of Pentium who were lagging behind in that area.
Now that the tables have turned quite a bit, and let's face it - everyone saw it coming since last April Core 2 Duo announcement he's desperately trying to save the sinking ship.
Hence, you gentlemen are simply feeding the troll. Notice how he doesn't respond to any of your comments? It's because he's nothing to respond with, he's a pathetic internet flame troll. That's what he's been reduced to.
repost and reword from my previous comment in the other thread just to respond to the funny claim below
abinstein said...
Penryn is just a Kentsfield shrink with 1) larger cache, 2) SSE4, 3) deeper power saving. Other than SSE4, the performance boost seems to be only 10~15% for 20% higher clockrate and 30% faster FSB speed. Such scaling is terrible. Not to mention the faster FSB will require a more power hungry north bridge, totally negating the better power saving of the CPU.
one trick that AMD fanbois would use would claim the IMC and the northbridge differences. Yes, the Northbridge would consume some energy, but in total the current Intel platform is still more power efficient. You do not need to claim the extra energy used in Northbridge or a particular type of ram in the server case; you just need to measure the power at the plug (power source)! Tech report has shown benchmarks with reference to the total power used, and Intel's Core platform is more efficient.
In other words, performance wise, in most cases Penryn @3.3GHz will still lose to K10 @2.6GHz, and take more power due to its need of high power north bridge.
you can claim whatever you want, just like AMD did in a presentation of Intel's platform power usage (by summing up all the max power of each components!). However, benchmarks have proven that Intel's Core platform is more power efficient.
Kudos to Intel for a stellar Q1. I’m more impressed than the analyst. Intel keeps raining on my parade. I consider myself pro AMD mostly because they’re the little engine that thinks it can.
One has to wonder where Sharikou is now.
Maybe he is in hiding as Intel is not slowing down at all.
IDF has just shown that Intel is far from going BK, far from slowing down, far from letting AMD catch up, let alone pass them again.
I said it before and I'll say it again. Intel got lazy. AMD took a great advantage of the situation, but now Intel isn't about to let something like that happen again.
Where has Sharikou gone!
"Anthony said...
Sharikou is becoming a fairly sorry looking individual, rather quickly. "
No - he looked that way from the beginneing.
This comment has been removed by the author.
Intel sales is down and profits are 'helped' by a tax benefit.
The Intel Dumb Forum is just a pure marketing exercise. They should be sued for calling it a 'developer' forum.
I know you Intel fanbois are releasing a sigh of relief, you can have your 2 seconds before the brutal assault by AMD.
Folks of the Intel PR pack - lex,roborat,tim,evil ...it is really impossible for sane people to support evil like Intel , you all might need professional medical help.
If that is not the case, you guys still have time to dumb your Intel stock ...lol
Abinstein is right.
Penryn is just a Kentsfield shrink with 1) larger cache, 2) SSE4 (significant for the video encoding improvements), 3) deeper power savings and 5) the faster FSB for bandwidth reliant apps like Cinebench
Why was Penryn compared against lower clock speed (3.3GHz vs 2.9GHz)?
Penryn: 45 nm, 1333 MHz FSB, 6 MB L2 Cache, 3.33 GHz
Core 2: 65 nm, 1066 MHz FSB, 4 MB L2 Cache, 2.93 GHz
Why the XQ6800 wasn't overclocked at 3333MHz???
Intel is an Illusionist!
Intel is playing with this bunch of kids mind.
It is the same chip and where is the difference of high-K + metal gate transistors?
Barcelona will kick the butt of Penryn!
LOL
P.S.
Just one week away!
haha, netrama ! You're a joke, AMD's assault? What kind of acid are you on?
We have yet to see Barcelona performance NUMBERS AND SPESIFICATIONS.
Perhaps Barcelona is delayed until late Q3? Would'nt surprise me.
LeeCooper said...
...
Why was Penryn compared against lower clock speed (3.3GHz vs 2.9GHz)?
Because it can. They are comparing the product that base on market frequencies. According to Intel, one of the official freq for Penryn is 3.33GHz
Why the XQ6800 wasn't overclocked at 3333MHz???
They are comparing the CPU performance, not IPC. Having high IPC is good for performance, having high frequency is good for performance too. What is more important is that Intel is able to do that within the thermal envelop.
What about overclocking the Barcelona to 3.33GHz to compare to Yorkfield then? some questions that ppl need to ask are that
1) what will be the power envelop for the Barcelona in 3.33Ghz? (IIRC, 95W @ 2.5GHz)
2) will it ever able to run stably in 3.33Ghz?
netrama
"it is really impossible for sane people to support evil like Intel"
But still people are supporting vastly bigger evil force that is dominating everything and even has huge influence on governments.
That insanely powerful company is called microsoft. If you thought Intel has bad business practices then MS is pure evil.
pointer
"IIRC, 95W @ 2.5GHz"
It was 120W at 2.5GHz.
"Not to mention the faster FSB will require a more power hungry north bridge"
We all know this is nonsense. Penryn will work in the same chipsets that the current Core 2 Duos can work in. For instance, the Intel P965 chipset.
Penryn will wipe the floor with Barcelona. Introduced at speeds in excess of 3.33Ghz. The architectural enchancements that Intel have made will mean that Penryn is more efficient than Barcelona, while still being clocked far higher.
My guess is people are getting smarter more often.
That's right. People realise that Intel is the "smarter choice", providing great performance whilst using less power. AMD is finished. They dropped prices on their dual core CPUs to well below $100 and still no one wants them.
Pat Gelsinger at IDF also talked more about Tigerton, the successor to Tulsa. Taking the core micro-architecture to MP servers and wiping out AMD's one remaining glimmer of hope. AMD will be forced to slash Opteron prices to $250 for the top end model.
It would be interesting to see if AMD can even walk out of 2007 in one piece.
Intel allowed Anandtech to do some limited testing on a Penryn system at IDF! http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2972&p=1
Penryn is much further along than Barcelona. I doubt we shall see Barcelona before 2008. AMD has a recent history of delaying products. R600 has been delayed so many times it's crazy.
:) Ha, ha, ha!
Please, Pointer,
what are you talking about.
Penryn is the same shrink like core 2 duo!
I want to see what is the real improvement, not higher clock speed!
Do you understand that??
And please you don't have to explain to me that you are a child.
Barcelona will go over 3Ghz too, you will see it!
Lol
AMD's roadmap shows Barcelona going to 2.6Ghz next year. That's not 3Ghz.
Intel, OTOH, has stated that Penryn will be introduced at "clockspeeds in excess of 3Ghz".
AMD BK Q2'08. Maybe sooner if AMD doesn't get more cash from somewhere.
leecooper
"Penryn is the same shrink like core 2 duo!"
Could you elaborate?
"Barcelona will go over 3Ghz too, you will see it!"
I have no doubt it will. If lucky it will happen on 65nm before 2009.
Though isn't it interesting that first Intel quadcore clocked at 2.66GHz last November and AMD hopes to get its highest end quads up to 2.5GHz by the end of this year. There have been rumours that Intel will release Penryn quadcores at up to 3.4Ghz this year. I wouldn't be surprised to see 3GHz quadcore Barcelona fighting against 4GHz Penryn quadcore.
Netrama - learn to type in clear concise sentences first and maybe you might write an intelligent response one day. Your engrish is terrible....
Holy crap, the AMD fanbois live in as much as a dreamworld as the Apple folks do.
Look I will make it simple, I don't care what goes in my servers (and we have thousands of them), as long as:
1.) they perform well
2.) They keey BTU's as low as possible
3.) They keep watt/hours as low as possible
4.) They keep costs as low as possible.
Right now, Intel has the lead. Firmly.
The AMD fanboi's can talk about Barcelona all they want, but they seem to think Intel isn't moving forward. Sure Barcelona MAY be faster than today's Core2 chips from Intel, but as I said before, when Barcelona hits some time in 2008, Intel will be there with something considerably faster than the CURRENT Core2 stuff.
Wake up folks, Intel isn't playing child like games here.
I swear the fanboi's claiming Penryn is just a die shrink must be skipping over all the other details that have been released about Penryn at IDF. Gordan Moore himself said that the move to 45um production is made possible by the use of high-k and metal transistors
Ho Ho said...
Could you elaborate?
Elaborate yourself:
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2972
Ho Ho and others Int...
Tell me why Intel need Penryn if they already have the fastest commercial cpu on the market?
Do you want to know why?
Because they are all affraid of what will Barcelona do to them!
Lol
Actually LeePooper - Intel announced a while ago that they would be making advances every year - one year enhancements to the current generation(ie penryn this year) and the next a new archetecture (ie nehalam (sp?) next year) that is *why* they are doing it. It has nothing to do with being "afraid" of AMD.
Also - regardless of what you've linked, the simple fact that they are switching to High-K in the transistors means it is more than a simple shrink. That is a *significant* advancement. (More than IBM and AMD saying "us too us too!" when in reality they're at least a year out from even going to 45nm if not longer given their current financial status). The inclusion of SSE4 and the other enhancments preclude this from being a simple die shrink (ala brisbane for AMD's x2).
Penryn will have 40% gain over core 2? I doubt it, but even if it's true, who cares? Barcelona will have a 200% advantage over Core 2. Penryn is too little, too late.
"PENIX said...
Penryn will have 40% gain over core 2? I doubt it, but even if it's true, who cares? Barcelona will have a 200% advantage over Core 2. Penryn is too little, too late. "
PLEASE link valid benchmarks to this 200% bullshit. I'd love to see them. And I don't mean AMD's powerpoint presentations or their graphs from their sites. I mean REAL world, third party benchmarks.
What's that?
You can't? I know you can't moron, it's because there aren't any.
At least Intel has allowed, once again, third party's to see and touch the new parts - same as with the Core 2 release (oh and OMG they were pretty much on with their claims then too).
AMD has shown NOTHING. No one has seen Barcelona, no one has touched it. I expect this to do in Tunisa to be more of the same as we've seen.
The only thing late here is Barcelona.
:)hm, ok Tim I see that you are very smart guy, gay I don't know what but please explain to me why would I buy a chip if the next year it will be old, and 30% slower?
Tell me Tim you are an commercialist like Intel, do you think that the cpu price is comparable like the price of socks or undershirt? Tell me, I will need to change the cpu every year?
Intel is saying: Dodn't by this year, next year will be much more faster cpu!
Oh, really, Ok, I'll wait for next year.
They are killing themselves!
Lol
"LeeCooper said...
:)hm, ok Tim I see that you are very smart guy, gay I don't know what but please explain to me why would I buy a chip if the next year it will be old, and 30% slower?
Tell me Tim you are an commercialist like Intel, do you think that the cpu price is comparable like the price of socks or undershirt? Tell me, I will need to change the cpu every year?
Intel is saying: Dodn't by this year, next year will be much more faster cpu!
Oh, really, Ok, I'll wait for next year.
They are killing themselves!
Lol"
I don't make Intels strategy Intel does, and regardless of your denial, that strategy is working quite well. Their strategies, (or AMD's) however, do not drive my purchasing decsions any more than it should yours.
In terms of personal use - I buy what has the best price/performance ratio at the time I'm getting ready to upgrade based on the budget I've set for myself. I don't sit around waiting for the next greatest thing - that's a vicious cycle of frustration.
In terms of enterprise use - the project timeline dictates the purchase and follows similar guidlines to personal purchases in that we have a budget to work with and will base what we purchase based on what the best price/performance is within those budget guidelines.
Intel's, (nor AMD's for that matter), "future" release's do not drive any purchasing decision I make unless that decision is going to be made *after* the release and subsequent performance validations are available.
Not that hard of a concept there, junior.
tim said...
At least Intel has allowed, once again, third party's to see and touch the new parts - same as with the Core 2 release (oh and OMG they were pretty much on with their claims then too).
I've seen the penryn benchmarks, and they are not impressive. They upped the MHz and doubled the cache. There is no innovation in Intel's strategy.
:), hm
Not that hard of a concept there. Yes, very very hard.
Tell me why are you Intel senior?
What are you trying to proof?
Why are you and the others responding on all posts that somebody wrote against Intel?
Are you googles cursors that has the job to put us on the right path?
Are you a 'Seven Nation Army'?
Lol
To Penix - yet you base your claims about barcelona's superiority on....nothing because so far that's whats been shown. Nothing. I guess we'll see when REAL benchmarks come out on both sides. If you are dismissing high-k as not being innovative then you're not only an idiot (which is not in doubt) but blind as well.
I must add that your ardent support of Sharidouche who has been proven to be an outright liar (ph.d? please, my 16 year old son is more intelligent and coherent than this moron), just puts you in the same light as Sharidouche. People like you are born to be made fun of due to your gullibility.
To LeeCooper - I support both companies equally - out of 17 computers in my home over half are AMD based systems - my last main desktop before this one was an Athlon64 3700+. I favor the company with my $$$ that gives me the best bang for my $$$ at the time I upgrade - again not a difficult concept to consume.
My last upgrade cycle happened to be when C2D was the best performing proc for the money - and well ahead of AMD dropping their prices to stupid levels. I recently upgraded my HTPC and my wifes PC to x2 4600's 65w because for the price the performance couldn't be beat. (proc and mobo for $119 was just too good to pass up.) Even with Intel's upcoming price cuts next week the deal was just to good. :)
As to why I personally respond to some of these posts - a) I'm insanely bored at the moment, and b) The level of fanboyism that plunges into outright stupidity in terms of what some of these guys post here is just to easy to have fun with (see my response to Penix above).
Penix what a fanboy you are.
Intel's 45nm process gives additional performance clock-for-clock, when going from 65nm.
AMD's 65nm parts performed WORSE than its 90nm counterparts.
Penryn is impressive.
"AMD BK Q2'08. Maybe sooner if AMD doesn't get more cash from somewhere."
Don't worry, AMD is getting between $10,000,000,000 and $20,000,000,000 from Intel once they are found guilty of restraint-of-trade.
Or maybe you think Michael Dell and other PC execs will perjure themselves on the stand ?
Sharifraud Has Absconded
San Francisco, USA: The ultimate AMD fanboy, named Sharkou, or "The Doctor" has gone invisible, suddenly stopping the flood of the stinky puke he has come to be famous for. Some say that he has committed suicide. Other's say that they last saw him riding a donky in east Afganistan, where one of his cousins, Osama Bin Laden is believed to be hiding. It is not very clear at this point what him pushed over the edge. One school of thought say that Sharikou was always delusional, and suffered from schizophrenia. In his catatonic episodes, he was known to keep staring at the screen, without responding to sane comments from the extremely intelligent visitors of his blog. However, there are others who refuse to believe in this theory. Conspiracy theorists believe that, Sharikou has lost the faith in the AMD leadership, and has gone to seek help from his notorious cousin, Osama Bin Laden. The truth is bound to come out sooner or later. Stay tuned.
Fox News: Fair and Balanced!
"Sharikou is becoming a fairly sorry looking individual, rather quickly...
Hence, you gentlemen are simply feeding the troll. Notice how he doesn't respond to any of your comments? It's because he's nothing to respond with, he's a pathetic internet flame troll. That's what he's been reduced to."
I actually find all of this amusing but would like to add that since Intel fired Sharikou pHd they have been on a roll. It appears Intel was severely handicapped while he was on their payroll. As far as Sharikou not responding, he is, just under the assumed name Penix. Compare the writing styles and comments and you too will find the truth. ;-)
Barcelona will be paper launched in June, expect full disclosure shortly that this processor family will be delayed by a few weeks and the ramp up will not occur as planned. The performance numbers that will be presented next week are very specific in regards to two particular benchmarks, there is a very heated debate about displaying general performance numbers at this time as they are just not that impressive yet.
AMD will not hit a 50% run rate until late Q2 in 2008. The financial numbers tomorrow are worse than expected and additional product news is not going to be good.
Also, forget about the high-end R600 as the XTX version was canned yesterday. The power requirements were too great for the board suppliers to contend with at this time. The top card will be the XT with 512MB of memory and performance falls right under the 8800GTX.
Argue all you want, but AMD is going down a dangerous path with late product introductions and the cost containment measures will only make this worse.
http://www.hardocp.com/image.html?image=MTE3NjkwNDc2MnRoZ08zVllVNFJfMV8xX2wuanBn
Heres some benchmarks from intel regarding Penyrn, take them with a grain of salt, but I found it interesting that a DC Penyrn is faster @ divx encoding that a QC Kentsfield
Forgot to include these new benches as well from Anandtech
http://www.anandtech.com/cpuchipsets/intel/showdoc.aspx?i=2972&p=2
"You do not need to claim the extra energy used in Northbridge or a particular type of ram in the server case; you just need to measure the power at the plug (power source)!"
Yes, I was talking about system power usage. For the same amount of memory and the same video cards, an Intel E6600 takes about 20W more power than an AMD X2 4800+ at idle and 40W more under load. The E6600 performs better in most (but not all) cases, but is also twice more expensive (both CPU and MB).
Penryn will consume more power than E6600. Barcelona, however, will stay in the same power envelope as K8 X2.
Boy, the Intel FUDsters are foaming at the mouth right about now, aren't they?
pointer said...
What about overclocking the Barcelona to 3.33GHz to compare to Yorkfield then? some questions that ppl need to ask are that
1) what will be the power envelop for the Barcelona in 3.33Ghz? (IIRC, 95W @ 2.5GHz)
2) will it ever able to run stably in 3.33Ghz?
Sorry, guy, doesn't look like it will be too big of a stretch, according to this story!
Quote: Agena is up and running and we learned that it can be a really nice overclocker. At stock voltage you can overclock the 2.5 GHZ native quad core part to at least 3 GHz. If you increase the voltage you will go even further.
So, if it can get to at least 3 GHz at stock voltage, I don't think 3.33 GHz will be too difficult.
Later
Randy Allen said...
"Not to mention the faster FSB will require a more power hungry north bridge"
We all know this is nonsense. Penryn will work in the same chipsets that the current Core 2 Duos can work in. For instance, the Intel P965 chipset.
Sorry guy, but this doesn't look like it will be the case...
Read this article.
All of the processors are run on Intel D975XBX2 'Bad Axe 2' motherboards, but the Wolfdale and Yorkfield processors needed motherboard voltage modifications to run correctly. This is the sign that all current Intel D975XBX2 motherboards will not support Penryn as modifications are needed.
So, if you feel up to doing some impromptu motherboard surgery, it ,will be compatible, otherwise, you're probably out of luck. Keep that soldering iron handy!
I'm so sick of the blatant fanboyism in these comments. I have reviewed the IDF Beijing reports. I have reviewed the AnandTech report. Both are inconclusive, unimpressive, and can be attributed directly to an increase in cache and/or clockspeed.
AnandTech states "Penryn vs. Conroe improvements to be in the 5 - 10% range at minimum depending on the application." This is not even CLOSE to the 40% bullshit Intel claim. Intel is spouting their phony benchmarks and you fanboys are eating it up once again.
Barcelona is in position to blow Intel out of the water and you fanboys are too blind to even acknowledge it as a likely possibility. All I hear is the chanting of this "HIGH-K!" buzzword. High-k is about as innovative as MMX, Strained Silicon, Netburst and all those other bullshit buzzwords Intel has thrown out in the past.
PENIX said...
I'm so sick of the blatant fanboyism in these comments
So then what does that make you, anything conclusive to back up your statements that Barcelona is as sexy as your wet nurse is?
Regardless of the innovation or lack of at least intel is showing SOMETHING
PS it seems your starting to lose it, no one is forcing you to read this blog
Man all this touching in here could get out of line. I only touch my wife so leave me out. Seriously be smart if it wasn't for AMD do you really think you could get to "touch" an Intel cpu for the price they would charge you for it. Innovation/new technologies are great for all & keeps these blogs busy as hell too. My 2cents worth! Duck! Here it comes again.
"Penryn vs. Conroe improvements to be in the 5 - 10% range at minimum depending on the application"
What part of minimum is it you don't get?
Why are you so threatened by Intel??
"So, if you feel up to doing some impromptu motherboard surgery, it will be compatible, otherwise, you're probably out of luck. Keep that soldering iron handy!"
Also note that power consumption scale up in cubic with increase in both voltage and frequency.
That is, a north bridge with 1.3x higher FSB clock rate and (presumably) 1.2x higher voltage will take 1.3*1.44=1.872, or about 87% more power.
That's ~15W in additional to the (conservatively) 18W power usage on the chipset & FSB. Thus expect a 65W Penryn to use more power than a 95W Agena when MB is taken into account.
Ignoring that you are just plain wrong about power scaling, you are aware that AMD cpu's have a chipset also, correct?
As a matter of fact, most AMD motherboard chipsets actually have more chips than Intel motherboards.
So much for the power saving of an IMC...
Intel has Core 2 Duo. It rules. All AMD's processors a load of garbage. Who wants a 6000+ or 4x4 space heater?!
http://techreport.com/reviews/2007q2/core2-qx6800/cine-power-peak.gif
Check out those power consumption numbers! 4x4 doubles as a space heater for sure!!
Penix is a fool. He claims Intel's "40%" number is BS, then comes out of nowhere and says Barcelona will be 200% faster! Even AMD have just kept refering to 40% in a single FP application.
Intel's high-k 45nm technology is the greatest breakthrough in the last forty years! What does AMD have? A broken 65nm process that produces CPUs slower than the 90nm process! They also have an ancient 90nm process for spaceheaters, sorry, I mean CPUs.
Sun has has some excellent new Intel based servers on show at IDF, 4P Tigerton server was showed. Compare the awesome performance of this to AMD's ancient 90nm junk scrapheap CPUs.
Intel should find some trivial things to sue AMD over. This will tie them up in court with more expensive legal fees and will ensure that AMD's BK is even sooner than the projected Q2'08.
AMD's losing market share like crazy. AMD is bleeding money like crazy. AMD has lowered prices to such an extent that they're going to lose cash even faster, but still people refuse to buy AMD trash.
A mid-range Core 2 Duo E6600 will easily frag all AMD's processors in performance and energy efficiency.
http://techreport.com/reviews/2007q2/core2-qx6800/cine-power-peak.gif
Ati is under attack on all sides from Nvidia. Almost six months ago Nvidia introduced the world's first DX10 GPU, and now they have a whole product lineup of them. Exp
AMD BK Q2'08.
I see Penix is working into a frenzy. Good. Maybe his head will explode.
Remember - for a true fanatic like him it's perfectly ok to make up numbers for his arguments - but when presented with factual data his answer is to rant. Good job.
As a note - no where in my postings do I say Barcelona won't perform. I simply say that we've seen no valid numbers from anyone, except AMD. At least with Penyrn Intel allowed outside testing, al be it limited. What has AMD shown? Power Point slides with theortical numbers.....>yawn<
And none of that changes the fact that AMD's financials are in the toilet and going down fast than Sharidouche on Hectors jock. All while Intel meets expectations and posts a profit AND gained market share.
So hold on to those delusions of your Penix, we know it's all you have. When the real numbers come out we'll see - until then stop with the 200% BULLSHIT.
abinstein. Bearlake is 65nm. Penryn 45nm. Put 1 and 1 together. Why more voltage if boards today can increase FSB no voltage?
Intel E6600 takes about 20W more power than an AMD X2 4800+ at idle and 40W more under load.
http://www.techreport.com/reviews/2007q1/cpus/power-idle.gif
http://www.techreport.com/reviews/2007q1/cpus/cine-power-total-energy.gif
Penryn will consume more power than E6600.
Poor fanboy.
Randy Allen said...
Intel has Core 2 Duo. It rules. All AMD's processors a load of garbage. Who wants a 6000+ or 4x4 space heater?!
The same people who 6 months ago wanted a P4 or Pentium D "spaceheater". By the way the 6000+ performs on par with the E6700 so I hope you are not inferring that the E6700 is also a load of garbage.
Randy Allen foolishly said...
Intel's high-k 45nm technology is the greatest breakthrough in the last forty years!
I hope you were being sarcastic.
More Randy Allen wisdom...
Ati is under attack on all sides from Nvidia. Almost six months ago Nvidia introduced the world's first DX10 GPU, and now they have a whole product lineup of them. Exp
Now if they could only produce working video drivers...
Tim said...
As a note - no where in my postings do I say Barcelona won't perform. I simply say that we've seen no valid numbers from anyone, except AMD. At least with Penyrn Intel allowed outside testing, al be it limited. What has AMD shown? Power Point slides with theortical numbers.....>yawn<
Remember when Intel released the Conroe benchies and then no one bought an Intel chip for the next 6 months waiting for Conroe? Perhaps AMD doesn't want to duplicate that scenario.
6000+ is an ancient 90nm junkyard space heater with awful performance. Below that of the E6600. http://xbitlabs.com/images/cpu/dualcore-roundup/charts/average.png
Intel will have two 45nm fabs online this year, and another two next year. AMD will still be using 90nm to make CPUs in 2008!
Remember when Intel released the Conroe benchies and then no one bought an Intel chip for the next 6 months waiting for Conroe?
For sure. Intel just sold no processors during that time, right?
Now if they could only produce working video drivers...
Compared to who? Ati? With the god awful Catalyst Control Center that apparently needs to have FIVE running processes in the background?! Ati doesn't even matter anymore. 8800 GTX is twice as fast as any Ati GPU.
What did HardOCP have to say about the 8600 GTS?
AMD/ATI should be very worried at this point. NVIDIA just released their entire DX10 mainstream lineup and it delivers. Compared to ATI’s current generation the GeForce 8600 GTS kicks its butt. Compared to NVIDIA’s own last generation, GeForce 7 series, the GeForce 8600 GTS kicks its butt. We experienced the highest playable settings we have ever seen at this price range. The XFX GeForce 8600 GTS XXX Edition is one of the best gaming video cards we have ever used at $239.99. It provided the best gameplay experience for the money and has further performance potential for hardware enthusiasts.
If you are in the market for a mainstream video card that delivers great gaming, the GeForce 8600 GTS should be the only GPU on your list. It provides the best gaming performance for the price in today’s games, and has DirectX 10 support for tomorrow’s games.
AMD BK Q2'08.
Tech4Life said...
The same people who 6 months ago wanted a P4 or Pentium D "spaceheater". By the way the 6000+ performs on par with the E6700 so I hope you are not inferring that the E6700 is also a load of garbage.
No it doesn't, the X2 6000+ doesn't even beat an E6600, and it consumes twice the power. Heck, the X2 6000+ even uses more power than a QX6800!
http://xbitlabs.com/articles/cpu/display/dualcore-roundup_8.html
E6600 -> Entire X2 lineup
http://techreport.com/reviews/2007q2/core2-qx6800/index.x?pg=13
X2 6000+ power consumption -> Entire C2D/C2Q lineup
So at least AMD got the bigger charts in something... too bad it's in the 'lower is better' categories. LOL
You AMD fanboys crack me up.
This is AntiFanboy, debunking one fanboy FUD at a time. :)
Randy Allen said...
Intel has Core 2 Duo. It rules. All AMD's processors a load of garbage. Who wants a 6000+ or 4x4 space heater?!
You are an idiot , it is well established now that the Core 2 Duo is the biggest scam Intel has pulled ever so far. Ground up this processor is designed for benchmarks only , the rest happened - thanks to Intel's huge marketing gimmicks , bribing Anand , Tom and other a**kissers and the perception of speed
among idi*t Intel fanboys who work at the retailers. No day has passed since middle last year , which did not have a stupid press release from Intel. What do think happens at the IDF - more marketing BS.
Do you think these are the behaviours of a true Technology Company ??
netrama,
I think your behavior is consistent with that that of a diehard desperate AMD fanboy. It must suck to be you right now, it's like seeing your favorite sports team getting hammered match after match. I feel you for buddy. ;)
AntiFanboy out. :)
"tech4life said....
"randy allen said....
Intel's high-k 45nm technology is the greatest breakthrough in the last forty years!"
I hope you were being sarcastic."
Seeing as how without it shrinking to 45nm would be damn difficult and anything below that (32nm and further) would be next to impossible, I'd say it's pretty significant. Maybe not the greatest in 40 years but pretty damn big none the less.
The *only* reason Penix and all brush high-k off is because AMD doesn't have it. Why? Because they CAN'T do it for at least another two years - and again, given the financials from AMD, probably longer than that.
Myself I don't want to see AMD bankrupt (whoever coined BK should be slapped, it's assinine). AMD goes down we're back to $500+ CPU's and frankly I don't want to see that...outside of all that even if AMD were to go chapter 11, they'd still be there - leaner and probably meaner. But that much further behind.
"netrama said...
Randy Allen said...
Intel has Core 2 Duo. It rules. All AMD's processors a load of garbage. Who wants a 6000+ or 4x4 space heater?!
You are an idiot , it is well established now that the Core 2 Duo is the biggest scam Intel has pulled ever so far. Ground up this processor is designed for benchmarks only , the rest happened - thanks to Intel's huge marketing gimmicks , bribing Anand , Tom and other a**kissers and the perception of speed
among idi*t Intel fanboys who work at the retailers. No day has passed since middle last year , which did not have a stupid press release from Intel. What do think happens at the IDF - more marketing BS.
Do you think these are the behaviours of a true Technology Company ??"
That has to be the single-most stupid post I have ever read outside of Sharidouche's statements - congratulations, you are now officially an idiot...
Randy Allen wisely said...
Intel's high-k 45nm technology is the greatest breakthrough in the last forty years!
tech4life daintily said...
I hope you were being sarcastic.
Of course he was being sarcastic, you expect serious discourse, especially in response to Sharikou?
netrama, I feel sorry for you.
The best reason to buy the AM2-6000 is because it is cheap compared to e6800,e6700, or the e6600. For $239 plus a $50 motherboard with ddr2-800 you can outperform 99% of the intel line up in both price and performance.
The AMD AM2-6000 is now rated the best buy over all other intel processors.
Except for you intel fanboys who have thousands of dollars to throw away the average comsumer needs a good deal with excellant performance, that leaves out 99% of the intel solutions.
The AMD AM2-6000 is the right chip at the right time with the best price and performance per dollar both in the chip and the platform.
Intel has nothing to compete with the price performance of the AMD 6000+, nothing.
oneexpert wrote:
The best reason to buy the AM2-6000 is because it is cheap compared to e6800,e6700, or the e6600. For $239 plus a $50 motherboard with ddr2-800 you can outperform 99% of the intel line up in both price and performance.
The AMD AM2-6000 is now rated the best buy over all other intel processors.
Except for you intel fanboys who have thousands of dollars to throw away the average comsumer needs a good deal with excellant performance, that leaves out 99% of the intel solutions.
The AMD AM2-6000 is the right chip at the right time with the best price and performance per dollar both in the chip and the platform.
Intel has nothing to compete with the price performance of the AMD 6000+, nothing.
Beep! Beep! Beep! Beep!
Fanboy red alert! Fanboy red alert!
http://xbitlabs.com/articles/cpu/display/dualcore-roundup_8.html
After the April 22 pricecuts, the E6600 will be faster, cheaper and run MUCH cooler than your beloved X2 6000+ space heater/CPU combo. ;)
Once again, I am AntiFanboy, debunking fanboy FUD one at a time. :)
netrama:
You are an idiot , it is well established now that the Core 2 Duo is the biggest scam Intel has pulled ever so far. Ground up this processor is designed for benchmarks only , the rest happened - thanks to Intel's huge marketing gimmicks , bribing Anand , Tom and other a**kissers and the perception of speed
among idi*t Intel fanboys who work at the retailers. No day has passed since middle last year , which did not have a stupid press release from Intel. What do think happens at the IDF - more marketing BS.
Do you think these are the behaviours of a true Technology Company ??"
Sooo, you got anything that could be defined as proof or fact besides the talking points you read on AMDZone and this blog? Oh wait, that is gospel to you.
I mean, all those FPS and such must be fake. And that cache, its only a gimmick right?
What a moron.
"http://www.techreport.com/reviews/2007q1/cpus/power-idle.gif
http://www.techreport.com/reviews/2007q1/cpus/cine-power-total-energy.gif
"
I do not dispute their measurements, though IMO the use of 700W power supply is a mistake; most of the power consumption will be due to the low PSU efficiency.
My numbers are based on comparison between two custom-made E6600 and X2 4800+ (not 5000+), both using the same memory, harddrive, and video cards. I only measured the (direct) current going to the MB, so no PSU inefficiency was at play.
Also, I did not use Cinebench but prime95 to load the processors. I have no idea why they use Cinebench, too.
You simply made a big mistake in calling me a fanboi. I acutally own both systems, and perform my own measurements for my own business. I think those who do not do so but who believe whatever they see on-line are the real fanbois.
"http://xbitlabs.com/articles/cpu/display/dualcore-roundup_8.html
After the April 22 pricecuts, the E6600 will be faster, cheaper and run MUCH cooler than your beloved X2 6000+ space heater/CPU combo. ;)"
Gosh.. the "performance" chart is just pure bullshit. They simply get paid by Intel. The "average" is totally meaningless - it doesn't reflect average workload at all. The substitute of SYSMark by some custom-made office tasks for "not work in Vista" is laughable.
The more accurate comparison between E6600 and X2 6000+ is this: if you do media/3D processing, then go for C2D; if you do scientific or engineering, go for K8; if you do office applications, then go for the cheaper one. For games, it depends, but generally buy the cheaper one and spend the saving on a better video card.
Abinstein wrote
Gosh.. the "performance" chart is just pure bullshit. They simply get paid by Intel. The "average" is totally meaningless - it doesn't reflect average workload at all. The substitute of SYSMark by some custom-made office tasks for "not work in Vista" is laughable.
See, that's the only excuse you fanboys have. Reviewers paid off by Intel this, Enron that (LOL thanks for the laughs Penix). When faced with facts, you dismiss them.
I actually believe Xbitlabs to be a credible source of information, if you have ANY proof that they (or Anandtech or THG) are being paid off by Intel, then please, show it for everyone to see. Otherwise STFU and stop spreading FUD like our resident Doc.
That chart is just a summary of the benchmarks made into a 'performance index' form. It takes into account a wide range of workloads, and yes, while the X2 6000+ will beat the E6600 in certain applications, it doesn't take away the fact that the E6600 is the faster processor [b]overall.[/b]
I stand by my assessment that the E6600 is the faster processor, runs much cooler, and will end up costing slightly less after the April 22 pricecuts. Win, win, win. AMD has lost this round, accept it and move on.
stand by my assessment that the E6600 is the faster processor, runs much cooler, and will end up costing slightly less after the April 22 pricecuts. Win, win, win. AMD has lost this round, accept it and move on.
Indeed. The E6600 will be a better buy for consumers. You might even say it's the "smarter choice"!
"while the X2 6000+ will beat the E6600 in certain applications, it doesn't take away the fact that the E6600 is the faster processor [b]overall.[/b]"
The reason that you are (one of the) true fanboi is your use of the term "overall." There is no "overall" in terms of general purpose CPU performance. Simply none. The fact is, half of those "benchmark" applications are programs that 95% users never use. How can you call that an overall indication of computer (PC) performance?
Only fanboi would believe such crap.
"Only fanboi would believe such crap."
So says someone who posts religiously on AMDZONE. ROFL
The E6600 wins more benchmarks than the X2 6000+. It's simple.
Of course, cry to your buddies at AMDZone and cry about how benchmarks are 'optimised' for Intel. ;) ;)
abinstein said...
...
Yes, I was talking about system power usage. For the same amount of memory and the same video cards, an Intel E6600 takes about 20W more power than an AMD X2 4800+ at idle and 40W more under load. The E6600 performs better in most (but not all) cases, but is also twice more expensive (both CPU and MB).
where do you get this data? Tech report said the opposite. At idle, E6600 consumes 121W while 5000+ 65nm(can't find data for 4800+) consumes 123W. At peak E6600 consumes 163W while 5000+ consumes 187W. Over a particular program execution completion, E6600 consumes 8555J while 5000+ consumes 9381J!
http://www.techreport.com/reviews/2007q1/cpus/index.x?pg=13
Penryn will consume more power than E6600. Barcelona, however, will stay in the same power envelope as K8 X2.
well, I'm not sure whether you do not follow up on the latest tech trend or blinded by your AMD faith, Intel's 45nm process allows it to increase the frequency while maintaining the thermal envelop. You will be really disappointed when Wolfdale/Penryn is releasen because it will be in the same TDP as the Conroe/Merom's.
That is, a north bridge with 1.3x higher FSB clock rate and (presumably) 1.2x higher voltage will take 1.3*1.44=1.872, or about 87% more power.
i'm not sure which background you are from, I'd really wish the those electric calculation is that easy in my college time. And further more, do you really know what % of the FSB power consumption compare to the overall chip?
abinstein said...
"http://www.techreport.com/reviews/2007q1/cpus/power-idle.gif
http://www.techreport.com/reviews/2007q1/cpus/cine-power-total-energy.gif
"
I do not dispute their measurements, though IMO the use of 700W power supply is a mistake; most of the power consumption will be due to the low PSU efficiency.
wow!! i really don't know what to say about this. All test system is using the same power supply. If there is any PSU efficiency issue, it affects the Intel system as well. I'm quite surprised that you still able to spin this! read this link for the test setup: http://www.techreport.com/reviews/2007q1/cpus/index.x?pg=2
Also, all of our test systems were powered by OCZ GameXStream 700W power supply units.
If AMD wastes more energy on an iffecient PSU it isn't going to waste less than Intel with a more efficient one. Also very funny that you would pair a $50 mb with a high end chip, and even funnier that you before claim AMD is more efficient with a certain chip yet later compare with AMD Inefficient 6000.
Post a Comment
<< Home