Intel stuck with the hot and slow image
I talked to someone in a datacenter yesterday, and he complained that Intel Xeons are too hot, and AMD Opteron is fast and cool.
Now, guys, that's the reality.
The majority of the servers running today are Netburst ones and they will take Intel to road of BK. A few Woodcrests won't do any good, because at the 4P level, the Tulsas are even worse than Dempsey.
I suggest to the datacenter guy that they should charge higher price for those who use Intel.
BTW, Core 2 Duo is completely unsuitable for running Windows Vista, you get a dreaful score of 1.0 out of the box.
Athlon 64 X2 3800+ PIB is being sold at $130 now. The x2 3600+ is now below $100.
posted by Sharikou, Ph. D. @ 12/15/2006 04:18:00 PM
38 Comments:
Could that be because the graphics driver wasn't installed? Don't you find it a little strange that the system jumped in score 3.4 (higher than the AMD system!) as soon as the correct driver was installed!?
With a low 3.0 score all crappy AMD products are unsuitable for Windows Vista.
You truly are an idiot. Your butt buddy in a 10 server "datacenter" doesn't mean jack shit.
Oh, and for the record, moron, Vista's perf score isn't just based on CPU.
...and now that I've looked at the artical you linked the suspicion that you are a true douchebag is confirmed...
If you're going to post 1/4 truths you phony it may be a good idea to NOT link your source...intelligent people will actually read it...
But since you're a Ph.d I'm *sure* your considered that...unless Ph.d stands for Pretty Happy Dimwit.
INTEL is kicking some major A$$
Selling all that pentiumIV from dirt cheap amortized 90nm factories. Got 65nm cranking on high margin Core2 that win every benchmark and more even against the latest and greatest 65nm cores.
Barcelona not on the scene in any volume till 2nd half of 2007. By that time the world will have been floored by 45nm Penrym.
Did you see that sorry 65nm paper from IBM/AMD at IEDM this year?
And what % of the Server market is 4P again?
Just curious...
BTW - you are an idiot - did you even read the last paragraph of the Reg article?
'But the upshot of our tests is that AMD is whistling if it thinks that Intel can't run Windows Vista in all of its Aero glory on G965 graphics. It can, and it looks great and it is our opinion that Intel is fully ready for 2007'
Damn, you are really lazy....
LOL!!! you forgot to add this....
"It was time to install Windows Vista RC2 on the Intel PC, and initially it looked dreadful scoring 1.0 on the Windows Experience Index. This was on the RC2 Microsoft driver 6.0.5472.5 so we downloaded and installed Intel driver 7.14.10.111 which made things much better. We were now able to raise the monitor refresh from 60Hz to 85Hz and the Windows Experience Index score shot up to 3.4, yet there is clearly more to come as this driver doesn’t have support for Shader Model 3.0, even though we know the hardware is capable of it."
YUP!! INTEL BEAT THE AMD MODEL 3.4 to 3.0!!!!
It's the graphics driver's fault that the Vista score is low.
It may sound as though we fudged those figures but we ran Windows Vista on our high-end Core 2 Quad test system with a GeForce 7950 GT graphics card and it too scored 1.0 on the Windows Experience Index. We updated the driver to version 96.85 and the score leapt to 5.0, which is pretty darn impressive.
You're really grasping at straws now.
BTW, how's those AMD 65nm chips coming? It seems they aren't much more power efficient than comparatively priced Core 2 Duo processors, which happen to be faster too. That's if you can find any 65nm chips from the few OEMs that might have them before the year is out.
Really? My E6600 got a 5.something when I tested Vista RC1.
You are an imbecile.
You shameless liar. Core 2 Duo only got a 1.0 because the drivers for the video card weren't installed.
You're such a a pathetic sack of sod. Your pathetic attempts at FUD are so transparently boring that they're really not even amusing anymore.
Did you even read that article? The 1.0 was caused by a buggy video card drive; they updated the driver and the score shot up to 5.0.
"Now, guys, that's the reality."
Did you know there will be a C2 based cluster built by BMW with 512 hot-swappable CPU's? Now that is reality too.
http://www.theinquirer.net/default.aspx?article=36367
"The x2 3600+ is now below $100"
Too bad you can't get a 65nm one anywhere for quite some time. I thought AMD produced a whole lot of them but for some reason only OEM's get them. Any idea why? Could it possibly be that AMD still can't produce enough of those?
I don't want to beat a dead horse or kick a man when he's down, so I won't have to cut and paste what everyone else has pointed out in your poor attempt at making something out of nothing at all.
And while I'm not much for name calling... just know that I'm thinking you're all the names everyone else has already pointed out and will continue to think those things until you man up and admit your mistakes.
Intel isn't all that. I run an 4600+ X2 and vista business edition RTM (not beta or RC).
My vista experience score is 5.0 for the proc, and about 5.5 or 5.7 for some of the other components. This is running vista 64bit. Intel will probably cry out even harder once the true 64bit tests come out on the true 64bit platform of vista. Intel fanboys, it's coming, just wait.
sharikou,
it has been a long time since you used the word "frag". I kind of miss it
Core2 is a absolute failure in real world work loads, especially in server applications.
Core2 when coupled to the really crappy intel platforms, use too much power and do too little actual work.
Core2 cannot be scaled.
Core2 has very limited memory bandwidth.
Core2 quad has such a slow front side bus after it divides to 4 cores it is a joke.
All major servers are switching to AMD opterons to gain peak though put and low power consumption.
In 99% of the server benchmarks conroes fail to perform and are always beat out by AMD opterons.
Intel is stuck with a hot slow image because they have earned it daily.
Intels platforms are antiques as are the pentium 3 conroe core2s.
Yet another clumsy lie.
I predict you delete this post with its comments within the next 24 hours, as you do with so many of your clumsy lies.
"I talked to someone in a datacenter yesterday, and he complained that Intel Xeons are too hot, and AMD Opteron is fast and cool."
Yeah, and I spoke to Pat Gelsinger yesterday and he says the new Xeons are fast and cool, and that Opterons are slow and hot.
LOL
don't worry, AMD just beat intel in creating a SLOW and HOT image with the 4x4.
I can't recall anything more humiliating than being beaten in performance and at the same time drawing twice the power. even the netburst didn't achieve that. GOOD JOB AMD!!
found this in http://computing-intensive.blogspot.com/
The Secret of MegaTasking - Revealed!
Just had a chance to talk to Henry from the A company, to understand the funky terms - MegaTasking ...
Q: What is multitasking?
A: The running of two or more programs in one computer at the same time.
Q: Then what is the different between MegaTasking and multitasking?
A: MegaTasking has some similar sense with the multitasking, but it is more than just computing?
Q: Can you elaborate further?
A: Sure! .The MegaTasking is about convergence. While our competitor is/was talking about the computing, networking and communication convergence, we aim much higher. It converge almost everything for your daily life stuff. For what you do in the living room, kitchen, laundry, neighbor, society, lan party, (342 words omitted). With our sophisticated design, advanced process, brilliant individuals, smart executives, enormous fanbois base, (1178 words omitted) ...
Q: So ...? What is it has anything to do with the living room? are you talking about your 'Live' stuff?
A: Nope, more than that.
Q: Is that so?
A: Yup, the moment you turn on the MegaTasking in a living room, you got a PC and heater.
Q: What if it is summer?
A: Err ... you just turn your living room to a Sauna Spa.
Q: ... then the kitchen?
A: With a proper casing, you just got yourself an oven. You can look at the screen for the recipe and bake the cake at the same time!
Q: ... ... then the laundry?
A: Just put your wet clothes close to the fan, it will instantly dry. Better than most of the commercially available clothes drier!
Q: ... ... ... then the neighbor?
A: What's more fun than directing the noise to your stupid neighbor that use our competitor's product? I'm sure our fanbois base would love this.
Q: ... ... ... ... and anything else?
A: What's even more fun than 'legally' disturb your opponent with noise and heat in lan party game competition?
Q: ... ... ... ... ... and some more?
A: The MegaTasking is a innovation for someone innovative. Think of it for few minutes, i'm sure you can list out more than what i have said.
Q: ... ok, anything else to say?
A: Yup, our MegaTasking is fast.
Q: How fast?
A: It can consume 1 MegaWatt in just 41 days consider full day usage. Our competitor not even close to that.
Q: What about the computing speed?
A: Sorry, I gotta take a whiz ... bye.
Whatever people may think of what Sharikou posts, he is merely giving Intel a taste of their own medicine,
so Intel Fanboys have no cause for complaint.Furthermore, what is important in graphics performance is not just frame rates but image quality, an area where Intel falls down.
See here: http://www.amdassets.com/comparison/
Quoted from article:
"But the upshot of our tests is that AMD is whistling if it thinks that Intel can't run Windows Vista in all of its Aero glory on G965 graphics. It can, and it looks great and it is our opinion that Intel is fully ready for 2007. ®"
so Intel Fanboys have no cause for complaint.Furthermore, what is important in graphics performance is not just frame rates but image quality, an area where Intel falls down.
See here:
People concerned about playing games don't use integrated graphics.
what is important in graphics performance is not just frame rates but image quality, an area where Intel falls down.
See here: http://www.amdassets.com/comparison/
Who would have thought that Intel intergrated graphics are slow!?? I thought they were quicker than a pair of Geforce 8800 GTXs in SLI!
Really this is just AMD saying "well, our processors are not as fast as Intel's. But when you compare an Intel CPU using an Intel IGP to an AMD CPU using an Nvidia IGP we're faster!
It's 140.29 USD at the link you provided. How did you arrive at 130 USD? In fact, this is the cheapest X2 on this site. At least on Newegg there's a part at 122USD...but nothing below 100USD.
Perhaps you're referring to the Sempron 3600+? Or are you making stuff up yet again?
http://sharikou180.blogspot.com
Pat Gelsinger yesterday and he says the new Xeons are fast and cool, and that Opterons are slow and hot.
You can trust Patty's opinion. He hates AMD.
I believe the only reason you give us links to your lies is to avoid legal trouble.
AMD has yet to demonstrate anything that beats Intel, I'm not holding my breath that it will happen anytime soon.
I can only imagine what you would be saying about FUDx4 if it came from Intel.
Hector could take a crap in your mouth and you would have great things to say about the taste and texture.
Core 2 Duo. One Core To Rule Them All.
..Yeah and you hate intel. Can I trust your opinion too?
You truly are an idiot. Your butt buddy in a 10 server "datacenter" doesn't mean jack shit.
I am sorry but I have seen Intel Xeons roast hard disks due to inadequate cooling. Those disk were running at temperatures 60 degrees centigrade and above, killing system stability and later themselves.
I heard of another case of a guy invested a significant sum of money to put a few Intel 1U servers in a few 1U slots of space in a datacenter to build a business. The whole thing literally went up in smoke. Had it been AMD processors in there, I dare say he could have succeeded.
Hot and slow?
That describes Sharikou PhD hatred of ( HOT )
and his computational IQ ( slow ).
You've reverted to more retarded subjects then even your BK and laptop fire articles.
LOL must be no bad news to find at all about INTEL..
Poor Sharikou.. you need to find a new hobby like getting laid more often buddy
I am sorry but I have seen Intel Xeons roast hard disks due to inadequate cooling. Those disk were running at temperatures 60 degrees centigrade and above, killing system stability and later themselves.
I heard of another case of a guy invested a significant sum of money to put a few Intel 1U servers in a few 1U slots of space in a datacenter to build a business. The whole thing literally went up in smoke. Had it been AMD processors in there, I dare say he could have succeeded.
Whose fault is that? Keeping hard drives cool is not Intel's problem, it's the system integrator's problems. N00bs... Idiots.
It looks like there are a few corrections that need to be made. Tulsa is not worse than Dempsey. Tulsa is in fact a pretty decent upgrade to the Prescott cored Xeon line. If Opteron didn't connect via HT, Tulsa would be a frontline processor. Also, heat is not as much a factor with Xeon since Intel moved to 65nm.
C2D is not a bad processor for 4-way. Woodcrest could easily run 4-way today and give AMD a real challenge if Intel had a quad FSB chipset. And, 4-way Clovertown would also be a serious contender to Opteron 8-way. It is a shame that Intel won't have a quad FSB chipset until a quarter after AMD delivers K8L. And, since the chipset is the limiting factor 45nm won't give that much of a boost to the server hardware in late 2007.
K8L will actually be out in volume in Q2 07. There is a least one supercomputer that uses K8L that has to be online by mid 2007; this can't happen unless the chips are available months earlier.
I also wouldn't claim superiority in 45nm process for Intel just yet. I've seen some addendums to the IBM/AMD process and it looks competitive.
I really wouldn't be surprised if Revision G only shows small improvements. When AMD does a die shrink, the transistors are not shrunk fully. In other words, Revision G is not really 65nm; it's probably more like 78nm. With K8L, AMD should have a more process optimized transistor and this should move closer to 65nm size.
A 512 cpu cluster may be news for C2D but this has been off-the-shelf technology for Opteron for awhile now. Anyone could buy one today from Opteronics. Opteronics is considered a low end HPC system. Cray has two levels higher than this with a mid level HPC system based on Octigabay technology and a high level HPC system based on Red Storm technology.
I'm still puzzled why some people keep insisting that 4X4 is hot. The best that Intel could do to hit back at 4X4 would be to design its own dual socket enthusiast system. If they design this around the fastest Clovertown the power draw would be similar to 4X4 since these chips are rated at 120 watts TDP. Naturally, this would give Intel the lead until AMD releases a K8L version of FX.
"C2D is not a bad processor for 4-way. Woodcrest could easily run 4-way today and give AMD a real challenge if Intel had a quad FSB chipset."
Core 2 will face real problems to scale beyond 2-way (4 cores), due to its lack of NUMA capability. Core 2 is intrinsically a dual-core design: both cores share one L2, which offers fast shared memory (multithreaded) performance.
Going to 2-way 4-core, there are two L2 caches to sync. The north bridge needs to be 2x faster (if core-to-core communication is about the same as core-to-memory one) to handle it. Going to 4-way 8-core, there are four L2 caches to sync. In theory, the demand on north bridge becomes (4,3)=6x more than the dual-core case.
IMO, it quickly becomes hard if possible for Core 2 to perform well on multithreaded programs beyond 4 threads/cores; it will need either a super-powerful FSB (one that scales n^2 with number of n cores) or a NUMA-like architecture.
Of course, people will argue that the number of programs to utilize more than 4 threads is few anyway, so it really doesn't matter (for now). Nevertheless, one thing is for sure: the claim from Intel's executives that the company is increasing # of cores aggresively (32 by 2010?) is either without any sincerety or with total ignorance.
"When AMD does a die shrink, the transistors are not shrunk fully. In other words, Revision G is not really 65nm; it's probably more like 78nm."
The "65nm" really means the "design rules of 65nm." You don't get to specify 78nm on a poly line; you specify a lambda. It is then up to the lithography to translate the lambdas to actual lengths in a way to optimize yield/capacity.
While it's possible today for AMD to make strictly 65nm-feature sized processors, they might actually obtain better total output by spacing the lines say 78nm apart. This allows AMD to maintain high yield while gradually shrinking the actual feature size. The immediate benefit of 65nm however is then its lower power consumption (which is likely to be improved along the way, too).
Such continuous-scaling will benefit Intel, too, and to some (lesser) degree it must be doing so, too. The difference however is that whereas such practice is desirable for Intel to have, it is necessary for AMD to survive. The latter simply has no spare fab nor money to copy-exact.
Whose fault is that? Keeping hard drives cool is not Intel's problem, it's the system integrator's problems. N00bs... Idiots.
ROTFL. Please tell that to all these Intel believers in Hong Kong that treat Tom's and Anand's as the authorities and who still think that AMD = HOT and Intel = COOL and so they choose to put these furnaces into a 1U box. I sure love it when an Intel fanboy calls another Intel fanboy NOOb and Idiot.
I, for one, will not even put an Intel furnace in a 2U box since that will box will never be stable even with Intel server boards unless you can find a data center that maintains 10 degrees centigrade whereas AMD processors on Tyan boards are rock solid.
This may all change now but I really wonder. I just wait for the next Intel sucker to get burned.
"Revision G is not really 65nm; it's probably more like 78nm."
I agree with this and think this is just a marketing way for AMD to save face ("ship 65nm before end of 2006"). They will do the same thing on 45nm as well. When Intel announces/releases the technology node it is both scaled to the ITRS target dimensions and has the perfomance they claim for the technology- it appears as though AMD has done neither with their "65" nm process (transistor performance with CTI is not at final technology target) It will likely not be until the end of 2007 until BOTH of these occur - but from PR perspective it certainly sounds better to be only 1 year behind (who cares about the little details like actual transistor performance and critical dimensions). What this means to Joe consumer is some modest improvements in the early "65" nm chips but don't expect OC capability (and also fmax scaling) until later in the "65"nm lifetime
I believe the general industry agreement on technology node is the 1/2 pitch at metal 1 layer (someone please correct me if I'm wrong here) - would be interesting to see what happens when someone cracks a chip open and actually measures this. This would also be somewhat consistant with the fact that AMD is only getting ~33% die size reduction on X2 Athlons at this point.
Does anyone else find it odd that AMD has not refined the "release" date of K8l beyond "mid-2007"
I mean to most this would mean anytime between beg of Q2-end of Q3 (~6 month window). As we are only ~3 months from the start of that window how is it that AMD cannot narrow the release date down further? (Maybe as a start to within 1 calendar quarter?)
I mean if they are really "releasing" at beg Q2 as some have siggested, the actual wafers on the final (initial) stepping would need to be started in Jan to allow for movement through the fab (~12-13weeks) and sort/packaging (?? weeks). I honestly can't see any significant volume (>20% of production) of K-late until Q4'07.
I'm waiting for all of the fanboy responses on this would be a reasonable ramp for AMD server only ...the same folks who are hammering Intel for only 25% Core2 shipment (across server, desktop, and mobile families) at end of year (which is <6 months after the product was released)
It'll be interesting to compare the K8l ramp rate to the Core 2 rate in the server area in terms of % production to previous gen product...
"Whose fault is that? Keeping hard drives cool is not Intel's problem, it's the system integrator's problems. N00bs... Idiots.
ROTFL. Please tell that to all these Intel believers in Hong Kong that treat Tom's and Anand's as the authorities and who still think that AMD = HOT and Intel = COOL and so they choose to put these furnaces into a 1U box. I sure love it when an Intel fanboy calls another Intel fanboy NOOb and Idiot.
I, for one, will not even put an Intel furnace in a 2U box since that will box will never be stable even with Intel server boards unless you can find a data center that maintains 10 degrees centigrade whereas AMD processors on Tyan boards are rock solid.
This may all change now but I really wonder. I just wait for the next Intel sucker to get burned. "
I run a datacenter with hundreds of Intel (intermixed with a few Opteron based) based servers, 1U and 2U and some older 3U. Stacked in 42U racks. Intermixed with massive amounts of storage. Roughly 28 Racks, nearly all are full you do the math...
I have *never* *EVER* had issues with cooling in the 8 years I've run this center. Why? Because my datacenter's cooling and airflow were designed properly. When you stand behind my racks the majority of the heat flowing out come from the racks with storage in them.
You are either very ignorant or very gullible (I'm sure you believe that Intel procs caused laptop batteies to explode too along with Sharidouche.)
People like you and Sharikook give real, true, IT professionals bad names.
I run a datacenter with hundreds of Intel (intermixed with a few Opteron based) based servers, 1U and 2U and some older 3U. Stacked in 42U racks. Intermixed with massive amounts of storage. Roughly 28 Racks, nearly all are full you do the math...
Let's see. What are in those 1U boxes you speak of? Intel Pentiums? Pentium IIIs? These guys are pretty safe in 1U boxes.
I have *never* *EVER* had issues with cooling in the 8 years I've run this center. Why? Because my datacenter's cooling and airflow were designed properly. When you stand behind my racks the majority of the heat flowing out come from the racks with storage in them.
Duh. Who said anything about the box going up in smoke due to inadequate data centre cooling? Oh the 10 degree centigrade statement. That was a joke about having to help the BOX'S cooling cope with its furnace of an Intel 2.8Ghz P4 architecture Xeon. XO in the US maintains 20 degrees centigrade in their data centres IIRC. Hong Kong data centres are much cooler at 15 degrees centigrade...just like everywhere else in Hong Kong where there is 'air-conditioning'...the temperature is set too low. Anyway, back to XO. No problems for Opteron boxes whether in 1U or 2U cases and for their disks. Rock solid. However, for 2U boxes with 6 disks in the front of the case that have Intel Xeon furnaces inside, the Xeons made the box so warm that the hard drives did not get any cooling which is, of course, not the fault of the data centre. My chums (the compnay is based in Hong Kong) asked XO staff for temperature readings right in front of the affected boxes (older Pentium III boxes had no problems...only the new Xeon boxes) which registered 25 degree centigrade. So they concluded that the rack needed more cooling (ROTFL) and that the boxes were fine. I had previously told them that the boxes were suffering disk problems because of heat issues which none of them believed and which I could not convinced them of since they had then switched to FreeBSD on those boxes which does not have any SMART support. When one box was freed later for another purpose, I asked for Linux to be installed so that I could prove the hot hard drives case. I could not believe it...hard drives running at temperatrues up to 65 degrees centigrade. No wonder they were sending disks back and forth to the vendor. The problem was later mitigated with additional fans in the cases to move more air. No extra cooling on the rack required of course. I hereby rest my case that Intel XEON boxes in 2U cases are trouble which can be confirmed by others on the Centos mailing list. All rack mountable cases I have seen all have the drives in the front except for the Sun X4600 which has loads of them in the middle. So anything that the vendor in the US that the company gets it boxes from supplies is very likely to have front mountable disks. Yet the Intel Xeons managed to roast those disks and probably even raise the temperature in the front of the case.
You are either very ignorant or very gullible (I'm sure you believe that Intel procs caused laptop batteies to explode too along with Sharidouche.)
Call me what ever you like. One of my pet peeves with my chums in that company was that they did not listen to me when I warned them that using a 3ware 7508 card in RAID5 mode is a performance disaster (they ended up switching to individual mirrors) and when I told that Linux is stable with good hardware and that therefore switching to FreeBSD on those problem Xeon boxes was completely unnecessary. FreeBSD 4.x was more stable than the Linux distros running 2.4 kernels because it pushes hardware less hard.
People like you and Sharikook give real, true, IT professionals bad names.
You just run a data centre. You have never been in an environment where you had to push hardware to its limits because you won't be getting any new hardware. I worked as a MTA administrator and I have been in the trenches fighting spam for over three years at the cost of my own family life since I was regularly called at night (HKT) to deal with issues for the company that owned the Xeon boxes above I have done 16 hour work days for months at times trying to mitigate email queues before getting a software solution (replace sendmail with postfix) and I don't need extra crap from people like you who don't listen to other people's opinion based on experience because that is exactly what I got from my chums then. I know from experience that Linux is stable on good hardware and I had seen boxes crash due to disk heat problems. So besides the issues from the regular bounce floods and email backlog, I had to also deal with the issues arising from my chums not following my advice to not use RAID5 on 3ware 7508 cards and to not switch to FreeBSD on the problem Xeon boxes which only hid the disk heat issues for another six months.
I was right on the hardware problems but I was in the minority on that position and not taken seriously maybe because I do not have tertiary education like the rest of my chums. I do not know what Sharikou has and I do not care. If I give real, true IT professionals bad names then it is the fault of those 'real, true IT professionals' like you who believe in theories and not in tests. Just like those scientists who believe in theories and who call laws of science 'theories' so that they can justify their position on their pet theories.
Enjoy running your data centre. Most XO data centre guys were pretty dumb IT professionals who only know how to handle a Windows interface. They hardly knew that those boxes had cooling issues nor what they contained just like you.
Post a Comment
<< Home