Here's a little teaser....

[dave_graham]

Not capitulating to my requests is an interesting way of responding to me just asking for some numbers. Numbers and not words is what XS is based on.

As for your time, you have spent a long time posting on this forum but cannot find the time to run a simple 30 second pifast bench?

Rather than reply to this post just run it and publish the results. Thanks.

Regards

Andy

i ran wprime, truecrypt, wrar.

you can see the results right here:
http://www.xtremesystems.org/forums/...&postcount=375

and

http://www.xtremesystems.org/forums/...&postcount=364

feel free to read those at your leisure.

i'll put pifast in my queue but, it's not top priority.

dave

[dogsx2]

and it isn't hard in the Intel section? c'mon...this is the whole Democrats-Republican back and forth banter that happens all the time.

relax.

dave

Are you sure??? Intel is

[Kayin]

Alright, this is funny. Skewering a person cause he hasn't run what is most likely the world's most useless benchmark? Give him a break. They're still trying to get stuff to RUN this, let alone run a waste of time program. I'd rather see some real world stuff-LAME, DivX, etc. Photoshop filters, that's one I use EVERY FREAKING DAY, (emphasis for the fact that I can't seem to find a way to not fire it up every day... it may signal addiction, I'm not sure)but something that showcases some real world performance where it counts.

Dave, I don't care if you wait till the Judgement Trump, take care of what is important and all this will fall in line behind.

And, since it seems to be so thin in this area, thanks for taking your time to try and provide us with information. I appreciate your work, and look forward to more in the future.

[informal]

Alright, this is funny. Skewering a person cause he hasn't run what is most likely the world's most useless benchmark? Give him a break. They're still trying to get stuff to RUN this, let alone run a waste of time program. I'd rather see some real world stuff-LAME, DivX, etc. Photoshop filters, that's one I use EVERY FREAKING DAY, (emphasis for the fact that I can't seem to find a way to not fire it up every day... it may signal addiction, I'm not sure)but something that showcases some real world performance where it counts.

Dave, I don't care if you wait till the Judgement Trump, take care of what is important and all this will fall in line behind.

And, since it seems to be so thin in this area, thanks for taking your time to try and provide us with information. I appreciate your work, and look forward to more in the future.

Well said man.My thoughts exactly

[GenTarkin]

i ran wprime, truecrypt, wrar.

you can see the results right here:
http://www.xtremesystems.org/forums/...&postcount=375

and

http://www.xtremesystems.org/forums/...&postcount=364

feel free to read those at your leisure.

i'll put pifast in my queue but, it's not top priority.

dave

can you do that bench on the 32M w/ 2 threads?
cuz with 2 threads my Core Duo and 2GB 533mhz laptop @ 2.16ghz scores 39s (yes core duo not core 2 duo)

[dave_graham]

can you do that bench on the 32M w/ 2 threads?
cuz with 2 threads my Core Duo and 2GB 533mhz laptop @ 2.16ghz scores 39s (yes core duo not core 2 duo)

32m on wiprime w/2 threads @ 1.9ghz = 41.906 seconds.

dave

[GenTarkin]

32m on wiprime w/2 threads @ 1.9ghz = 41.906 seconds.

dave

hrm....I dont know if thats good for barcelona then if a cpu clocked 260mhz more wins by almost 3 seconds....wonder what the core 2 would do @ 2.16ghz
or OVERALL lol wonder what the barcelonay would do @ 2.16ghz how close it would be.

[s7e9h3n]

hrm....I dont know if thats good for barcelona then if a cpu clocked 260mhz more wins by almost 3 seconds....wonder what the core 2 would do @ 2.16ghz
or OVERALL lol wonder what the barcelonay would do @ 2.16ghz how close it would be.

Since I'm waiting on another mobo for my cpu's, I decided to see what my 2218's could do @ a comparable speed to Dave's 2347's in Wprime. Although the Barcelonas do show a bit of performance increase from K8, it's quite obvious that our current bioses are crippling the quads. The mobo makers who are claiming support for quad AMD's have quite a bit of work to do.....

under windows xp pro, sp2, on a 3992 (single plane) with 2 x 2347 B1 processors and 4 x 1GB DDR2-667 (CAS 5) ECC/Reg modules:

4 total threads

32m: 23.015 seconds
1024m: 742.875

My Opties @ as close as I could get them to the 2347's:




BTW, thanks for all the suggestions to get my quads running correctly but I've decided to let AMD figure out what's wrong with the Tyan setup and not make it my own problem. Should be up and running again with a fully functional system by early next week.....

[KTE]

Thanks Steven. Just as a further comparison:

Opteron 2210 1.8GHz 4GB 667MHz Win x64 = 26.844/856.171 sec (belongs to justapost)


(although this is a desktop part, 4 cores, and far higher frequency/tight timings)

Q6600 3150MHz 4GB 4 threads = 13.37 sec (belongs to ...cant remember )

[doompc]

E6300 (2MB L2)
@ 7x266 DDR2-800 = 50.203


@ 7x333 DDR2-667 = 40.219


@ 7x400 DDR2-800 = 33.407

[STEvil]

can you do that bench on the 32M w/ 2 threads?
cuz with 2 threads my Core Duo and 2GB 533mhz laptop @ 2.16ghz scores 39s (yes core duo not core 2 duo)

That a really fast Core Duo. Most Core 2 Duo T7400 are around 40-42 seconds according to this, and looks like Core Duo/Core 2 Duo are pretty well equal clock for clock in it anyways: http://forum.notebookreview.com/showthread.php?t=123570

T7400ES (2.16ghz 667fsb, 2x1gb 533mhz ram 4-4-4-12, 945GM.)
XP Pro-32
32M : 40.020s
1024M: 1273.53s

2x 5120ES @ 2.8ghz (7x400 5-5-5-x 4x1GB FB-DIMM's)
XP-64
32M : 15.875s
1024M: 494.782s

[KTE]

That a really fast Core Duo. Most Core 2 Duo T7400 are around 40-42 seconds according to this, and looks like Core Duo/Core 2 Duo are pretty well equal clock for clock in it anyways: http://forum.notebookreview.com/showthread.php?t=123570

T7400ES (2.16ghz 667fsb, 2x1gb 533mhz ram 4-4-4-12, 945GM.)
XP Pro-32
32M : 40.020s
1024M: 1273.53s

I have a C2D lappy lying around somewhere here. I think its either that or T5600.... or maybe higher.
I'll check it out soon and try getting some wPrime scores, although it has no OS on it yet and hasn't been touched for quite a long time.

[GenTarkin]

I just ran it again with nothing going on in the background..or everything I could rid of as far as chat windows =P
and got this:

38.687 sec on the 32M one
UPDATE: after few more runs was able to attain 38.276 sec on 32M

thats with my Core Duo 2.16ghz and RAM @ 533mhz - laptop

Im running Vista Ultimate 32bit

UPDATE: Just completed the 1024M w/ same system. Result: 1236.75 sec

[tiro_uspsss]

I have a C2D lappy lying around somewhere here.

thats so mean! no fair! i dont even have a lappy - but if i did i sure as hell would know where it is

[duploxxx]

looks like i missed some fun here:

anyhow here is a cpu-z of one of the first EVT samples.



@s7e9h3n nice to see you have some samples to play with. But Bios does magic in these barcelona figures and behaviour! fixed bios for oem was only released last week. can't you get a 790 board somewhere?

[prostoman]

looks like i missed some fun here:

anyhow here is a cpu-z of one of the first EVT samples.



@s7e9h3n nice to see you have some samples to play with. But Bios does magic in these barcelona figures and behaviour! fixed bios for oem was only released last week. can't you get a 790 board somewhere?

Wow dude, what board are you using to run that Phenom? Is it AMD's 790 reference board?

[duploxxx]

Wow dude, what board are you using to run that Phenom? Is it AMD's 790 reference board?

from cpu-z 1.40.5 beta it is able to recognize it is a barcelona. 1.40 wasn't able at all.

it's just a wrong screen from cpu-z about the phenom part.

i had a 2s system tyan system btw for a few weeks from second week of august.

[generics_user]

after reading thorugh these thread i get the feeling that s7 is running on a x1 ht link multiplier (according to CPU-Z), maybe that's the problem with these benches

i can't try it myself because the only 2s system i have is the ones of my dad and this is a first gen 940 board running 2 265 cpus , but all the screens of K8 cpus show the correct ht link speed...

and s7 s DC optys also show the correct ht link ( do they run in the same board as the quads???)

[prostoman]

from cpu-z 1.40.5 beta it is able to recognize it is a barcelona. 1.40 wasn't able at all.

it's just a wrong screen from cpu-z about the phenom part.

i had a 2s system tyan system btw for a few weeks from second week of august.

Oh, I see it now. I was blinded by the AMD Phenom sign

[tsuehpsyde]

Well I just found out we have a pair of Barcelonas at work (2347s) and have this entire week and noone told me. But when I found out last night, I asked about them and they tried booting them on our current Tyan motherboard we had for the 2P Opteron platform. However, it won't boot (or wouldn't last night, anyway). The product guy should be in today, and I'm going to see if I can help him get it booted with at least one proc. I'll post the motherboard we have when I get a chance, but again, not sure what it is at this point.

Also, I think we have 0721s while s7 has 0725s so we probably have B0s I'm guessing. I now feel s7's pain of having two procs to try out and not being able to. Argh.

[flippin_waffles]

I know this is probably the wrong place to post this, but it's thread got locked for some reason (maybe it was generating more interest that the Intel/Havok news....) Anyway, looks like AMD is going to produce a tri-core. I wonder if it will be native. One article says it will be.

http://www.hardtecs4u.com/?id=1189807351,51761,ht4u.php

translated

And from the Inq

http://uk.theinquirer.net/?article=42369

[JohannesRS]

That was amazing.

People think that everybody has what, 40 hour in a day to work for free for them and give them results? Give all us a break.

Dave, take your time and bring us the results when they come.

I think that everybody already know that for the server world there are ALWAYS bios issues, usually called "imaturity of the plattaform". They are always in there, doesn't matter the manufacturer. Intel has those too, and the prove is quite easy to find: it took more than 6 months after release for their server parts "core2 like" start to harm the opterons selling, despite the fact that the core2 arquiteture always were faster than the at time opterons.

why that? no chipset worked at time.

So give everybodyu a break.

But, anyway: Dave, could you explain to us what exactly makes a server part bios so harder to clean than a desktop one? It's a fact... but I still can't see the reason behind it.

And thanks a lot for all info and time!

[tsuehpsyde]

An update for what it's worth. The motherboard we're using is a Supermicro H8DME (but the POST screen says H8DM8, oddly enough). Either way, we flashed it with the newest BIOS from 9/7/07 and now it works. Old BIOS was 11/28/06, so there's a clue to why it wouldn't even turn on before. Got a set of 2347s with 16GB of PC5300 ECC in there, and they're now building out the server to install an OS on it and start testing it. As far as working properly, we'll have to see but the BIOS is counting 8 cores. I took pics, but they're on the phone, so I'll have to get them off when I get home tonight if I can figure out how.

Just letting s7 and Dave know in case they want to know of another board that works. If the machine gets setup and I have access to it, I'll run what tests I can, but we'll see what they're doing with it first.

[s7e9h3n]

l
@s7e9h3n nice to see you have some samples to play with. But Bios does magic in these barcelona figures and behaviour! fixed bios for oem was only released last week. can't you get a 790 board somewhere?

A rd790 board shouldn't be a problem for me to get, BUT a 2s rd790 board is impossible to land....

after reading thorugh these thread i get the feeling that s7 is running on a x1 ht link multiplier (according to CPU-Z), maybe that's the problem with these benches

i can't try it myself because the only 2s system i have is the ones of my dad and this is a first gen 940 board running 2 265 cpus , but all the screens of K8 cpus show the correct ht link speed...

and s7 s DC optys also show the correct ht link ( do they run in the same board as the quads???)

CPU-Z is giving a bugged reading and the DC opties are running in my Asus L1N64

Well I just found out we have a pair of Barcelonas at work (2347s) and have this entire week and noone told me. But when I found out last night, I asked about them and they tried booting them on our current Tyan motherboard we had for the 2P Opteron platform. However, it won't boot (or wouldn't last night, anyway). The product guy should be in today, and I'm going to see if I can help him get it booted with at least one proc. I'll post the motherboard we have when I get a chance, but again, not sure what it is at this point.

Also, I think we have 0721s while s7 has 0725s so we probably have B0s I'm guessing. I now feel s7's pain of having two procs to try out and not being able to. Argh.

If the bios is from Tyan, you're going to have some major issues booting the board. The Tyan I used was running a custom bios from AMD which [allegedly ] allowed the cpu's to function properly.

[s7e9h3n]

An update for what it's worth. The motherboard we're using is a Supermicro H8DME (but the POST screen says H8DM8, oddly enough). Either way, we flashed it with the newest BIOS from 9/7/07 and now it works. Old BIOS was 11/28/06, so there's a clue to why it wouldn't even turn on before. Got a set of 2347s with 16GB of PC5300 ECC in there, and they're now building out the server to install an OS on it and start testing it. As far as working properly, we'll have to see but the BIOS is counting 8 cores. I took pics, but they're on the phone, so I'll have to get them off when I get home tonight if I can figure out how.

Just letting s7 and Dave know in case they want to know of another board that works. If the machine gets setup and I have access to it, I'll run what tests I can, but we'll see what they're doing with it first.

Yep, although AMD's preferred Tyan as their main launch partner for the last few years, it seems as if they're having more compatibility problems than Supermicro this time around