First "real" Nocona vs. Opteron review?

[DonPMitchell]

Extracting facts from bad information is just part of life in the field of computers. Either you are reading a benchmark made by someone with a financial conflict of interest, or by someone with a powerful religious point of view. You're reading Microsoft marketing or you're reading Linux/GNU dogma. Either way, people like that bend the truth, its become a deeply engrained part of both cultures, and so we are all up to our necks in propaganda.

I think its safe to assume that benchmark measurements are honest. At worst, measurements that show the "wrong thing" are omitted. It just means we have to work a little to try to see the light.

 

[ectx]

Originally posted by: mamisano

Originally posted by: rck01
Honest debate isn't working so you resort to personal attacks. Typical fanboy behavior - and a waste of my time.

Disengaging...

RCK

Well, that's an easy way out of answering NUMEROUS unanswered questions...


come on RCK, you can do it...

I do like to hear more from RCK (for real). I do hope he would post the link when his report comes out. I don't have time and patience to digest all the points but healthy debates are the keys to future improvents.

 

[DrMrLordX]

I'd like to know why he only tested one Opteron box that was only using Opteron 248s. What, no 250s? You even had 3.6 ghz Noconas in there. Sheesh!

 

[Zebo]

Originally posted by: DonPMitchell
Extracting facts from bad information is just part of life in the field of computers. Either you are reading a benchmark made by someone with a financial conflict of interest, or by someone with a powerful religious point of view. You're reading Microsoft marketing or you're reading Linux/GNU dogma. Either way, people like that bend the truth, its become a deeply engrained part of both cultures, and so we are all up to our necks in propaganda.

I think its safe to assume that benchmark measurements are honest. At worst, measurements that show the "wrong thing" are omitted. It just means we have to work a little to try to see the light.

Nah. You don't have to work that hard don mitchell, aticles anand/toms and half a dozen others (expect anything to do with monitors since it's so subjective) are dead accurate which is why they match and why they are trusted and read by millions everyday. Is it a scientific journal? Nope but closer than any of the afore mentioned sites because the financial incentive is readership loyalty not sales and margins of those sales.

 

[matman326]

If you want to see the biggest problem with Intel's Xeon compare 4 way or 8 way systems. The Opteron will acctually scale higher the more cpus you add due to thier dedicated memory and due to thier dedicated "link" to the northbridge. An Intel system maxes out the FSB at a 4 way system. Now with the move to an 800Mhz FSB they might make it scale into the 4 way but don't even think about an 8 way system. This is where AMD trumps Intel every time. Don't even try to dispute that.

 

[mamisano]

Originally posted by: DrMrLordX
I'd like to know why he only tested one Opteron box that was only using Opteron 248s. What, no 250s? You even had 3.6 ghz Noconas in there. Sheesh!

The Intellistation-A only comes with the 248 as the top processor.

My opinion is that in the Workstation environment, AMD is seriously lagging in Chipset features. In the Server environment it is not necessary to have all the bells and whistles, but on the Desktop it is crucial, especially when trying to compete with the incumbent Xeon.

 

[mamisano]

Well, RCK1's review is up at Inforworld.... what a JOKE!

http://www.infoworld.com/article/04/08/13/33TCworkstation_1.html

They have NO benchmarks listed.... Just some babbling about how the Opteron sucks a Multi tasking. I guess we need to take their word for it....

 

[Sunner]

Originally posted by: mamisano
Well, RCK1's review is up at Inforworld.... what a JOKE!

http://www.infoworld.com/article/04/08/13/33TCworkstation_1.html

They have NO benchmarks listed.... Just some babbling about how the Opteron sucks a Multi tasking. I guess we need to take their word for it....

Yep, that's not so much a review as it's a rant.

 

[Falloutboy]

LOL that review has to be the worse I've ever sean. maybe if you would back your findings up....maybe with some benchmarks.

 

[sandorski]

Wow, that settles it then. He says it is, it must be, right? I pity anyone who uses that "review" to draw facts from.

 

[justly]

I was keeping an eye out for rck01?s review also, I can honestly say good job ??



To mamisano for posting a link to it first.

All I can say to rck01 is that you have FAILED to provide an answer to many of the honest questions asked of you in this thread. It appears you have also FAILED to give the readers of InfoWorld an unbiased and complete picture by not specifying at the level of multitasking that causes the Opteron to lose to the Xeon (is it 2 treads 3,10,50,10000). Also your comment about more memory makes for better multitasking is only partially true, while it is true that to little memory will hinder performance (be it multitasking or not), but that does not mean by simply increasing the Xeon system memory to 4 GB (to match the Opteron system) will make it a better multitasking machine than it already is. The fact is that since the Xeon system you tested did out perform the Opteron system in your tests only goes to show that your test suite did not require more than 1 GB or the Xeon would have had to deal with more disk swapping and the penalty for that would surely have made the Opteron look much better than it did.
My comments only scratch the surface of why your review has FAILED for all but a select few that may use their workstations EXACTLY as you have tested. While your results may be beneficial to those select few I think it could be very misleading to anyone else that doesn?t fit your testing suite to a tee.

 

[Markfw]

Originally posted by: Sunner

Originally posted by: mamisano
Well, RCK1's review is up at Inforworld.... what a JOKE!

http://www.infoworld.com/article/04/08/13/33TCworkstation_1.html

They have NO benchmarks listed.... Just some babbling about how the Opteron sucks a Multi tasking. I guess we need to take their word for it....

Yep, that's not so much a review as it's a rant.

And he says " the 2.4GHz Opterons show their limitations and the A Pro starts to crawl." in the review (if you can call it that), but the 248 is 2.2 ghz, and on IBM's website, they also say "AMD OpteronTM Series 200 processors up to 2.2GHz with 1MB L2 cache, dual-capable". This guy is a moron and has no clue how to test computers, or maybe even use one. BS on the multitasking slowdowns, and I know that firsthand.

If you can't even get your facts straight, how could you possibly do a review anybody could believe ? Based on my personal experience, I think this moron could be outright LYING about the workstation performance.

 

[DAPUNISHER]

Propaganda is a powerful tool, and so is the guy who used it :laugh:

 

[RoosterKooster]

IBM ought to have his ass. That "review" is nothing but a piece of "journalistic" libel.

 

[justly]

Just for fun can anyone configure an IBM IntelliStation A Pro as tested?

Here is the link to IBMs IntelliStation A Pro (view all models page), and below are some quotes from the article that describe the IBM system used.

?With its 4GB of memory and Nvidia 1100 graphics card, the IBM A Pro should have been the ultimate workstation in this review.?

?The NF 600 sports two 3.2GHz Xeons, a gig of memory, and some very fast SATA hard disks. This matched nicely with IBM's SATA drives, although the IBM A Pro had 4GB of memory. ?

?Cost:
$8,345 as tested

Platforms:
Windows 2000 and XP Professional, Red Hat Enterprise Linux?

 

[bbbl67]

Originally posted by: rck01
Typical. Your reaction to a poor showing by your CPU of preference is to dismiss the test as being irrelevant. A bit pathological, don't you think?

Ri-i-ight. Okay, I'm a little curious about the testsuite product that you used in the test, *HTP* Analyzer (i.e. Hyperthreading Analyzer). According to its product write-up at http://analyzer.csaresearch.com/, it runs a test script which is "a classical linear benchmark", like ZD, Sysmark, Bapco, etc. And in order to load it down you use a "concurrent workload simulator". These simulators are Database, Workflow, and Multimedia, as you call it. It seems to me that these are synthetic benchmarks, as they aren't real database, workflow or multimedia workloads, but rather simulations of them. The only real application benchmark running is the single-threaded test script, while the other threads run synthetic benchmarks to slow the system down in the background. These simultaneous workloads are *pretending* to be database, workflow or multimedia workloads.

Since this is a Hyperthreading analyzer, it would stand to reason that this program knows how to detect and use Hyperthreading priorities. It would also stand to reason that a processor without Hyperthreading would be at a disadvantage.

A Hyperthreading processor would have two threads, a high-priority thread A and a low-priority thread B. Thread A will always get more priority than Thread B. You can put the real-world application test script on the high-priority A thread, while putting the synthetic workloads on the low priority B thread, and because A thread gets more priority than B thread, you can easily starve the synthetic workload simulations from getting any real CPU time, since when push comes to shove, Thread A is always king.

A non-Hyperthreading processor on the other hand would just use standard Windows task-switching priorities and give equal time to all threads, agnostic about whether they are real-world or synthetic.

Let me ask you something, does this HTP Analyzer measure how much time it took the synthetic workloads to execute in the background, or does it only measure the foreground test-script times?

Unfortunately, in my position I don't have the luxury of becoming emotionally attached to products.

Interesting that you mentioned the word "luxury".

 

[Accord99]

Originally posted by: bbbl67

Since this is a Hyperthreading analyzer, it would stand to reason that this program knows how to detect and use Hyperthreading priorities. It would also stand to reason that a processor without Hyperthreading would be at a disadvantage.

A Hyperthreading processor would have two threads, a high-priority thread A and a low-priority thread B. Thread A will always get more priority than Thread B. You can put the real-world application test script on the high-priority A thread, while putting the synthetic workloads on the low priority B thread, and because A thread gets more priority than B thread, you can easily starve the synthetic workload simulations from getting any real CPU time, since when push comes to shove, Thread A is always king.

A non-Hyperthreading processor on the other hand would just use standard Windows task-switching priorities and give equal time to all threads, agnostic about whether they are real-world or synthetic.

Intel's implementation of HT does not work like that. Both threads will get basically the same amount of wall time in the CPU, regardless of the priority setting. What you describe would happen with the non-HT processor, if the synthetic thread is assigned a lower priority than the real-world thread. The main benefit of the HT processor is that running two CPU intensive threads in general results in a combined throughput higher than running just one.

 

[RoosterKooster]

I find no recent evidence Randall C. Kennedy is anything but an Intel loyalist with obvious motives to continue his Hyperthreading crusade. I searched for his name with google and the trail of the last 2 years is obviously littered with his slanted, er, rooted with his tunnelling preachings that Hyperthreading is the only way to go. I won't post a bunch of links - there are many. His claims are outrageous. He touts his non-linear testing tool is superior to the various linear tests. The problem is his tests are engineered to promote Hyperthreading.

Infoworld should be embarrased by misinforming the IT community by relying on the drivel from this "out of gas" stoolie.

HT Crusader Spreads the Gospel

 

[sandorski]

Originally posted by: RoosterKooster
I find no recent evidence Randall C. Kennedy is anything but an Intel loyalist with obvious motives to continue his Hyperthreading crusade. I searched for his name with google and the trail of the last 2 years is obviously littered with his slanted, er, rooted with his tunnelling preachings that Hyperthreading is the only way to go. I won't post a bunch of links - there are many. His claims are outrageous. He touts his non-linear testing tool is superior to the various linear tests. The problem is his tests are engineered to promote Hyperthreading.

Infoworld should be embarrased by misinforming the IT community by relying on the drivel from this "out of gas" stoolie.

HT Crusader Spreads the Gospel

It's probably P4 Optomised to the extreme, which would likely explain why an Opteron might not fair well running it.

 

[bbbl67]

Originally posted by: Accord99
Intel's implementation of HT does not work like that. Both threads will get basically the same amount of wall time in the CPU, regardless of the priority setting. What you describe would happen with the non-HT processor, if the synthetic thread is assigned a lower priority than the real-world thread. The main benefit of the HT processor is that running two CPU intensive threads in general results in a combined throughput higher than running just one.

There is an implicit priority amongst the two logical processors (threads) of Hyperthreading. They will get the same amount of time as each other, provided there are no resource conflicts (eg. registers, memory locations, etc.) between them, and then the primary logical processor will always get first dibs on the resource. Just so long as the primary thread has control over a resource, then the second thread will always have to wait around.

 

[Accord99]

Originally posted by: bbbl67
There is an implicit priority amongst the two logical processors (threads) of Hyperthreading. They will get the same amount of time as each other, provided there are no resource conflicts (eg. registers, memory locations, etc.) between them, and then the primary logical processor will always get first dibs on the resource. Just so long as the primary thread has control over a resource, then the second thread will always have to wait around.

There is no primary or secondary logical processor and HT has no method of assigning greater value to one thread over another.

 

[splice]

InfoWorld LOL. That mag. is complete cr@p! I get that mag free several times a month. It goes directly in the Recycle bin.

 

[Noworkia]

Originally posted by: mamisano
Well, RCK1's review is up at Inforworld.... what a JOKE!

http://www.infoworld.com/article/04/08/13/33TCworkstation_1.html

They have NO benchmarks listed.... Just some babbling about how the Opteron sucks a Multi tasking. I guess we need to take their word for it....

That review sucked. It is incomplete. Where's the data? It should be www.lackofinfoworld.com