How about some large complex application test for Conroe?

[Acanthus]

Originally posted by: dmens

Originally posted by: Acanthus
The problem is their typical bench is 1m, which runs SPi entirely in cache and gives a rediculously unrealistic performance level.

It's 8MB, and the dataset in cache argument is meaningless. The fact that banias, dothan and yonah (with smaller caches) all did so well in superpi is indicative of superpi being an positive architectural outlier with the P-M uarch. So naturally it is extremely misleading to compare two uarchs using superpi only, or any other individual benchmark for that matter. I simply find it amusing to see people call superpi useless. All isolated benchmarks are useless.

That i would agree with, (isolated benchs) and i didnt know XS changed their "de facto standard" to 8MB, last i knew it was 1MB. 8MB and higher are more indicative of performance.

I just wish they would post stock benches and show the performance increases they get with all the tweaking. That would give us the best of both worlds, and it would get them a lot more hits as a website.

 

[Accord99]

Originally posted by: Acanthus
That i would agree with, (isolated benchs) and i didnt know XS changed their "de facto standard" to 8MB, last i knew it was 1MB. 8MB and higher are more indicative of performance.

The 8MB refers to the amount of memory SuperPi allocates for its 1 million digit test.

 

[Acanthus]

Originally posted by: Accord99

Originally posted by: Acanthus
That i would agree with, (isolated benchs) and i didnt know XS changed their "de facto standard" to 8MB, last i knew it was 1MB. 8MB and higher are more indicative of performance.

The 8MB refers to the amount of memory SuperPi allocates for its 1 million digit test.

Ahh, thank you for clarifying.

 

[Pr0phetX]

Originally posted by: Duvie
The biggest problem is the morons who have gotten early chips (cough* extreme *cough) are only interested in stroking their little e-penises for world records in superpi, 3dmock, pcmock, etc.....They dont even concentrate on something someone would actually run their computer for.....

Just wait for the real reviews and if in some great surprise it shows some sort of weakness in a complex app you use, you will have some A64 and AM2 benches to compare to....

NOthing hurts in waiting for competent reviews.....plus there is limited chipset support right now and it will only get better...

I wonder why your not a WR bencher? And why dont you bring your high and might words and post them at XS so the REAL experts can respond directly to you.

Ha.....and you want to be mod.

 

[Fox5]

Originally posted by: Pr0phetX

Originally posted by: Duvie
The biggest problem is the morons who have gotten early chips (cough* extreme *cough) are only interested in stroking their little e-penises for world records in superpi, 3dmock, pcmock, etc.....They dont even concentrate on something someone would actually run their computer for.....

Just wait for the real reviews and if in some great surprise it shows some sort of weakness in a complex app you use, you will have some A64 and AM2 benches to compare to....

NOthing hurts in waiting for competent reviews.....plus there is limited chipset support right now and it will only get better...

I wonder why your not a WR bencher? And why dont you bring your high and might words and post them at XS so the REAL experts can respond directly to you.

Ha.....and you want to be mod.

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

 

[DrMrLordX]

Originally posted by: Fox5

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

Bingo. Most WR benchers I've seen don't actually care about using the machine for anything important after it's overclocked; they just want to overclock for the sake of setting records.

Duvie's the kind of person who would want to put the machine to work 24/7 after getting it stable. He's too busy with CAD and other apps to set world records.

 

[Pr0phetX]

Originally posted by: DrMrLordX

Originally posted by: Fox5

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

Bingo. Most WR benchers I've seen don't actually care about using the machine for anything important after it's overclocked; they just want to overclock for the sake of setting records.

Duvie's the kind of person who would want to put the machine to work 24/7 after getting it stable. He's too busy with CAD and other apps to set world records.

Riight.

 

[Duvie]

Originally posted by: Pr0phetX

Originally posted by: DrMrLordX

Originally posted by: Fox5

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

Bingo. Most WR benchers I've seen don't actually care about using the machine for anything important after it's overclocked; they just want to overclock for the sake of setting records.

Duvie's the kind of person who would want to put the machine to work 24/7 after getting it stable. He's too busy with CAD and other apps to set world records.

Riight.

plus most will post speeds that are not remotely stable but will last long enough to run superpi 1mb....Often times many of the apps they benchmark I could run on an unstable system by quite amount....

I wont run benches and post them unless they are stable speeds...My testing suite has been the same for sometime and has always been rich with encoding (many), video editing (many), file archiving, CAD rendering (many), etc...

 

[Pr0phetX]

Originally posted by: Duvie

Originally posted by: Pr0phetX

Originally posted by: DrMrLordX

Originally posted by: Fox5

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

Bingo. Most WR benchers I've seen don't actually care about using the machine for anything important after it's overclocked; they just want to overclock for the sake of setting records.

Duvie's the kind of person who would want to put the machine to work 24/7 after getting it stable. He's too busy with CAD and other apps to set world records.

Riight.

plus most will post speeds that are not remotely stable but will last long enough to run superpi 1mb....Often times many of the apps they benchmark I could run on an unstable system by quite amount....

I wont run benches and post them unless they are stable speeds...My testing suite has been the same for sometime and has always been rich with encoding (many), video editing (many), file archiving, CAD rendering (many), etc...

Well why dont you put together a test for them to take. Is prime not enough nowadays to detect stability?

And i still dont understand why you haven't shown your face around XS to let all those "morons" understand. I mean you do know that some of those guys providing these tests work for the companies that you buy your components from right? These aren't little kids like overf at . Most of these guys that have the es chips are in their late 30's early 40's. They've actually "done" things for this community and made names for themselves.

All im really saying is that.......if your gonna talk so much crap. You might as well go to XS and practice what you preach.

 

[TekDemon]

Originally posted by: Pr0phetX

Originally posted by: Duvie

Originally posted by: Pr0phetX

Originally posted by: DrMrLordX

Originally posted by: Fox5

Um... what? It just sounds like he is looking for something other than what the typical World Record benchers look for. I always assumed that the WR benchers only focused on max clock speed and stability and ultimately didn't care about the superpi results, otherwise why would they bother ever using an architecture that doesn't get good superpi/3dmark results?

Bingo. Most WR benchers I've seen don't actually care about using the machine for anything important after it's overclocked; they just want to overclock for the sake of setting records.

Duvie's the kind of person who would want to put the machine to work 24/7 after getting it stable. He's too busy with CAD and other apps to set world records.

Riight.

plus most will post speeds that are not remotely stable but will last long enough to run superpi 1mb....Often times many of the apps they benchmark I could run on an unstable system by quite amount....

I wont run benches and post them unless they are stable speeds...My testing suite has been the same for sometime and has always been rich with encoding (many), video editing (many), file archiving, CAD rendering (many), etc...

Well why dont you put together a test for them to take. Is prime not enough nowadays to detect stability?

And i still dont understand why you haven't shown your face around XS to let all those "morons" understand. I mean you do know that some of those guys providing these tests work for the companies that you buy your components from right? These aren't little kids like overf at H. Most of these guys that have the es chips are in their late 30's early 40's. They've actually "done" things for this community and made names for themselves.

All im really saying is that.......if your gonna talk so much crap. You might as well go to XS and practice what you preach.

lol I totally agree...the people over at XS aren't exactly idiots who don't know what they're doing-especially the people who are setting records and have access to Conroe this early.

and I say this as a noob over at XS (which I joined mainly since I was bored and wanted to solder capacitors everywhere and see what it did haha).

EDIT: I edited the quote to fix the highlighting since Pr0phetX was trying to type the HardOCP H but the forum software read it as a highlighting tag

 

[DrMrLordX]

Just gotta say that I think Duvie has some good points. The problems inherant to synthetic benchmarks aside, just because "real world" benches need a reference to be useful doesn't make them useless. You'd think that these guys getting engineering samples would at least be able to do benchmarks with one other reference system. Sometimes they do this, sometimes they don't.

I've read Duvie's contributions here on Anandtech and I have to say that I see no reason for him to go elsewhere to contribute. He does pretty good work here when he has the time, and he's even teamed up with other AT members to verify results, etc.

Personally I don't care to bash people on other sites too much. What I would like to see is some of the more regular contributors here wind up with pre-release hardware to do some more comprehensive testing. Duvie or Avalon or Zap or . . . whoever.

 

[Absolute0]

The problem with the guys bashing my friends at Xtremesystems who have ES CPUs is twofold

1) you obviously haven't gone there much and aren't up to date on what's going on. It's more than super pi, it's almost every benchmark i know of and a few i don't

2) obviously, you don't CARE that your "real world" benches aren't used, because you don't register and post asking for them to try a new benchmark.

Here's the truth of the matter, they're in the industry, they have the ES chips, you don't, so whine and b1tch all day but they'll run whatever tests they want, and if you ask nicely they'll run what you tell them to run. Don't complain because they aren't setting up a bunch of AMD computers of completely matching designs. They are giving what they are because they have extra time and want to share. I dont think any of them are obligated to share anything. Be thankful they aren't gaming and folding all day and not releasing results, which they are perfectly entitled to do.

The fact that they are only running the benchmarks they are asked to run does not make them morons. If you guys had a Conroe you wouldn't get it to boot because you wouldn't have a motherboard. And after you talked to the guys at Xtremesystems, went through the channels and got a board, you probably wouldn't be able to overclock it. FUGGER, FreeCableGuy and the others are very smart, i know.

You got beef with the big guys, take it up with them, don't rabble rabble rabble to your friends at Anandtech about how big bad Xtremesystems guys won't share "real" benches.

 

[Vee]

I would interpret the OP question as a wish for knowledge about how Cornroe performs on real life working software. And that the OP is also aware of the problem that many benchmarks correlate poorly to real performance.

Even benchmarks that purport to represent real world performance and for that purpose are made up from segments of working applications can be put together in a way that is severely misleading. Here it has been a matter of carefully choosing tasks and applications that show off the favored architecture. And avoiding those that expose weaknesses.
(If you want a benchmark that correlates well to real performance, then IMO the choice should be Veritest Winstone.)

However, I would say that it was primarily the rather special Intel netburst architecture, the P4 and Prescott, that could be (and was) hyped out of proportion this way.
Conroe doesn't have the weaknesses that P4 have. So a diverse workload will not suddenly show a dramatically different picture.

What I'm saying is that I think we can expect common benchmarks to show a much more representative picture of the Conroe, vs AMD, than we are used to see from Pentium4. I believe the Conroe will prove to be as "honest" as AMD's cpus. The one exception is of course that the very large cache could concievably be intentionally targeted.

But there is thus not really any reason to doubt that this Intel desktop cpu, available in the future, will have a performance advantage, big enough to be characterized as "significant", on all AMD desktop cpus that are widely employed today.