For those who arent familiar with Mr Ou, he has a nasty habit of sniffing out and publically exposing BS...much to the chagrin of those who would pull the wool over our eyes.
In his article, Mr Ou points out:
Quote :
Nelson compared AMD’s Opteron 2350 2.0 GHz quad-core processor (may not ship again until Q2-2008 when the TLB bug hopefully gets fixed) to Intel’s older 65nm “Clovertown” E5335 and E5345 processor which were released in Q4 2006. These weren’t even the newest 65nm G Stepping Clovertown processors from mid-2007 with lower power consumption; these were the older stepping released in 2006. But Intel launched their latest 45nm “Harpertown” processors in November of 2007 and these chips were excluded from this “study” on AMD versus Intel energy efficiency. This is a classic case where the measurements are most likely accurate, but what’s being measured isn’t.
First let me say that I am not using this to bash AMD.....they didnt do the study. They may or may not have provided hardware for the study, thus skewing it, but it is moot for my purposes, and the real culprit here is Neal Nelson and Associates for publishing what is a blatently biased study.
Second, I call on Thunderman to find us a recent study to back up his claims, rather than a single "cooked" study, reguritated by The Inq and FUDzilla.
Finally, to Mr Ou....well done and thank you very much for continuing to expose BS where it rises..
---------------
Bonuses
+5 Genital Ambiguity/-25 Discouraging the removal of butt ugly god face penalty/-2 Blada/+5 Parental Wisdom/+1 Euphemism/+25 Too Often Quote Points/+5 Too Much Time On My Hands/+15 obscure quote/+ 2 Needless Repetition/+1 Stating the Obvious
Neal Nelson & Assoc. has always have the tendency to help out AMD in their studies. From the last three "studies" they published that I can remember, everyone of them favored AMD.
Now, while I'm not questioning Mr. Nelson's motive, it is definitely deceiving to,
1. Publish results based on completely different testing methodology (measuring total system power consumption, then claim power efficiency for a particular processor.)
2. Using outdated hardware for publication.
Although George Ou has been constantly criticized to be an Intel fanboy, but I always find his comments very insightful. I guess those who really has an idea of the background and the information are getting drowned by finger pointings.
--------------- Paul Otellini: "The initial ones are multi-chip, but so what?' You guys are misreading the market if you think people care what's in the package."
All of my recent test results have shown the for some configurations or in some ways Intel was better, and for other configurations or in other ways AMD was better. The white papers that we publish with our test results show this quite clearly.
The last three papers have shown that for some common configurations and workloads AMD servers are more energy efficient. We believe that this is because Intel has standardized on FB-RAM and FB-RAM uses more power than the DDR2 used by AMD.
The test machines were configured to have very similar specifications. 2.0/2.33 GHz Xeons and 2.0 GHz Opterons. Identical RAM sizes, disks, operating systems, tunables, databases and transaction loads. We will test with Intel's 45 nm parts but these cpus are advertised as having similar power usage to the 65 nm parts, and Intel continues to use FB-RAM, so I don't expect to see any major changes in energy efficiency on the these new tests. Neal
Do you have anything to refute George Ou's claims that you used outdated 65nm steppings from 2006 instead of the latest G0 steppings?
I can assure you 45nm will cut down on power dramatically compared to 65nm. I wouldn't judge a CPU purely by the 'advertised TDP' - surely a person of your position would realise that.
Is this any different to thunderman's posts? Hmm?? Not really ... just the opposite flamebait.
Well the OP is more literate ... the argument is simply the opposite.
Waiting for the horde to descend ...
Bit slow today ...
Next ... the comments about AMD lying ... ho hum ... it's getting old now guys.
Lets talk about the lies Intel told about the end of line Netburst Prescott CPU's eh??
130W TDP ... more like 145W ... burned out two mobos on the THG endurance test. It's on record If anyone cares to link it. The HSF's could not keep up either ... despite being boat anchors.
Lets talk about the RAM power draw on them Intel Servers too .... ouch !!!
Lets talk about scaling past a few sockets ... high end servers are still AMD only .... FSB is just too bandwidth limited.
Lets talk about paying companies not to buy AMD products.
All in all cranking the FSB up to 1600 should effectively squash the overclockers from getting more headroom out of the silicon tho.
Somebody should test the max the FSB can go without getting unusable. I mean there must be a wall somewhere. Anybody have the time to lower his cpu multiplier and see if he can find a wall to the FSB on a current board? It might be far tho so don't burn anything only for this.
Somebody should test the max the FSB can go without getting unusable. I mean there must be a wall somewhere. Anybody have the time to lower his cpu multiplier and see if he can find a wall to the FSB on a current board? It might be far tho so don't burn anything only for this.
On most quads the limit is between 450-500 MHz QDR. On Duals it's between 500-550ish MHz QDR. More on water.. and I've seen over 600 at times, but it's rare.
Edit: Btw, it doesn't kill overclocking. The current planned 1600 MT/s FSB processors are all EE's and have an unlocked multiplier. The rest of the processors are still running on the 1333 MT/s FSB.
Message edited by cnumartyr on 01-21-2008 at 02:33:51 PM
[quotemsg=
The test machines were configured to have very similar specifications. 2.0/2.33 GHz Xeons and 2.0 GHz Opterons. Identical RAM sizes, disks, operating systems, tunables, databases and transaction loads. We will test with Intel's 45 nm parts but these cpus are advertised as having similar power usage to the 65 nm parts, and Intel continues to use FB-RAM, so I don't expect to see any major changes in energy efficiency on the these new tests. Neal[/quotemsg]
Now, Let's assume that the only thing that matters is "Advertised" TDP.
Since we are talking about Power Usage based upon workload, you are still skewing the results.
Why Test a 2.33 Ghz part when a 3.0Ghz part has the same TDP?
Does your comment also seem to ignore general improved performance of the Penryn?
This alone should lead to a 1/3 increase in relative performance.
While there is no doubt some tests in which AMD servers excel, I use them for running VMWare ESX Servers, your statements do not show as somebody who is trying to forthright testing, but rather somebody who crafts a test to obtain the desired results.
It's best not to fool those with technical understanding.
#1) Yes, It is true that the memory for Intel servers require more power than those used in AMD Servers.
#2) Yes, It is true that Intel had released newer 65nm parts that cut power but you chose not to use those.
#3) Yes, It is true that Intel released newer 45nm parts that both cut power and increased performance, but you chose not to use those.
#4) Yes, It is try that Intel has processors nearly one-third faster with the same TDP but you chose to use the slowest CPUs so as to maximize the effect of the less efficient memory and minimize the effect of the more efficient CPU.
#5) Yes, you used AMD processors that are not generally available but failed to use the latest Intel processor that have been available for months.
Please Don't Bring arguements to the table that assume the readers here are ignorant.
As a whole, we are not.
Those statements may fly for those without sufficient technical knowledge, but here they seem to just reinforce your agenda.
---------------
If its good in theory but not in practice,
its not good theory.
Get a life people. Who cares?? If you feel the need to post or argue about cpu tdp, then you really have no life do you?
Says the person who has so little life they feel the need to point out to others who are debating topics they are passionate about to get a life..
It's a Hardware site, what do you expect? Do you think we are going to talk about cars or what we did over the weekend? If stuff like this bothers you.. you might want to consider leaving!
Get a life people. Who cares?? If you feel the need to post or argue about cpu tdp, then you really have no life do you?
Shut your cake-hole bytch. You seem to always have time to post your stupid rants about other peoples topics. Go back to your hole.
If people can find out lies or deception about article and news, which is about 90% of the media today then let it be told. It only benefits everyone as a consumer.
---------------
Q6600 G0 @3.6 1.34v | HD3870X2 1GB | 24" Westinghouse L2410NM| 19" 1907FP | Asus Maximus Formula X38 (Rampage Mod) | Kandalf VD4000 LCS | Ultra X2 750W | Supreme FXII 8.1 | Audio FX Pro 5.1 Headset | 4x1024 Ballistix Tracer 5-5-5-14 1066mhz
Seems like Neal is being called out by a few other people. His excuse for using the older intel chips seems pretty weak. Neal said, "The 65-nm (Intel) parts are being shipped in volume to customers today, so those are the parts that I was able to borrow to run these tests." While the 45nm chips are not shipping in volume, at least they are shipping, unlike the Barcelona chip that hasn't begin to be shipped. Which by the way he declined to say where he got it from...hmm.
Seems like Neal is being called out by a few other people. His excuse for using the older intel chips seems pretty weak. Neal said, "The 65-nm (Intel) parts are being shipped in volume to customers today, so those are the parts that I was able to borrow to run these tests." While the 45nm chips are not shipping in volume, at least they are shipping, unlike the Barcelona chip that hasn't begin to be shipped. Which by the way he declined to say where he got it from...hmm.
