K10 Scores starting to surface

Quote:

---

Originally Posted by MR_SmartAss

Both SuperPi and specFP_rate are not representing the real-world performance for server computing. Anyway, a CPU that performs good in SuperPi, also performs good in games.

---

+1:yepp:

I can't wait to see the full suite of benchies people will run on Barcelona so I can finally grab a quad, whether AMD or Intel:D

Quote:

---

Originally Posted by MR_SmartAss

Then why all Netburst sux in Super Pi, while K8 kicks the crap out of them? :rolleyes:

---

Give a Netburst 6GHz+ and tell me, who craps out who? :D ;)

Be wary of stuff like this..
http://img458.imageshack.us/img458/6606/ekafbz5.gif

http://img296.imageshack.us/img296/7...ona1ph6.th.jpghttp://img235.imageshack.us/img235/1...ona2rs4.th.jpghttp://img513.imageshack.us/img513/6...ona3ua6.th.jpg

Clearly the above is fake, just doing it to show you how wary you should be of benchmarks these next couple of days.

Quote:

---

Originally Posted by Jacky

no not necessarily, the improvement from "B00 chip from May to a B02" is probably due to some important parts being turned of in the B00 (maybe L3 or even L2) - the fact that *some parts were turned off* was already stated somewhere - thus it tells us nothing about the retail barcelona.
and most agree that coolaler's results show a B01 that is going to launch on sept. 10, however we don't know if the results aren't fake.
but don't forget that gary also said this about the results:


6 more days until we find out :)

---

My comment was solely dealing with the credibility of Gary Key, nothing else. He was a long time bencher here on XS and was offered a job over at Anand. The man is fully credible.

Personally I don't care if Phenom/Barcelona is a Core2duo destroyer or not, as long as it is decently competitive, and overclocks well. I am plainly a Amd fan but not a fanboy, and will choose their products over Intel's 90% of the time. I'm glad only server chips are out in September, allows time to ramp up the clocks for Phenom. I've stated in posts some time ago that Amd had serious issues with k10 process, and I'm hopeful these will be ironed out for the Phenom launch late 07/early 08...


Ply

Quote:

---

Originally Posted by MR_SmartAss

Both SuperPi and specFP_rate are not representing the real-world performance for server computing. Anyway, a CPU that performs good in SuperPi, also performs good in games.

---

But if this particular CPU also performs good in any other app, then there is no such strong relation Super Pi <-> games, but that this CPU is good at any kind of code.

Super Pi uses x87 code, depends on memory acesses (so a good cache + memory prefetchers will help) and has many loops, where Conroe's macro op fusion helps a lot.

Quote:

---

Originally Posted by Plywood99

My comment was solely dealing with the credibility of Gary Key, nothing else. He was a long time bencher here on XS and was offered a job over at Anand. The man is fully credible.

Personally I don't care if Phenom/Barcelona is a Core2duo destroyer or not, as long as it is decently competitive, and overclocks well. I am plainly a Amd fan but not a fanboy, and will choose their products over Intel's 90% of the time. I'm glad only server chips are out in September, allows time to ramp up the clocks for Phenom. I've stated in posts some time ago that Amd had serious issues with k10 process, and I'm hopeful these will be ironed out for the Phenom launch late 07/early 08...


Ply

---

QFT! How else can we solve all of the my Speculation is better than your Speculation crap? 5 more days, phew.

http://img528.imageshack.us/img528/8329/13sex6.gif

:eek:

if 13s @ 2.0 ghz is true (doubt it, hope so though), i'd rather see 2M/4M to be the norm. 1M would be the new 128k. click done.

Quote:

---

Originally Posted by Jacky

no not necessarily, the improvement from "B00 chip from May to a B02" is probably due to some important parts being turned of in the B00 (maybe L3 or even L2) - the fact that *some parts were turned off* was already stated somewhere - thus it tells us nothing about the retail barcelona.
and most agree that coolaler's results show a B01 that is going to launch on sept. 10, however we don't know if the results aren't fake.
but don't forget that gary also said this about the results:


6 more days until we find out :)

---

B1 will not be the launch core so coolater's results are invalid.

dave

Quote:

---

Originally Posted by biohead

if 13s @ 2.0 ghz is true (doubt it, hope so though), i'd rather see 2M/4M to be the norm. 1M would be the new 128k. click done.

---

Have faith mate......13s will be slow for these babies :)

Quote:

---

Originally Posted by MR_SmartAss

B1 has the CPU_ID registers with full CPU name and model and without the "Engineering Sample" or "ES" strings. Are you sure B1 is not the launch stepping?

---

yes. quite sure. the launch stepping is different and will have a different nom de guerre. Can't disclose the exact stepping version due to NDA restrictions but it's not B1.

cheers,

dave

SP33DFR34K, where did you find those images?

dave_graham, is it true that AMD's NDA include ritual suicide ? :D

no ritual suicide.

dave

Quote:

---

Originally Posted by dave_graham

yes. quite sure. the launch stepping is different and will have a different nom de guerre. Can't disclose the exact stepping version due to NDA restrictions but it's not B1.

cheers,

dave

---

Curious, care to venture a prediction on how a 2P Barcelona Opty 2350 (@ 2.0 GHz) might stack up against a 2P Opty 2224 (@ 3.2 GHz)?

Quote:

---

Originally Posted by doompc

SP33DFR34K, where did you find those images?

dave_graham, is it true that AMD's NDA include ritual suicide ? :D

---

Oh I made them to prove a point. Very easy to make REALLY believable images. So don't believe everything you hear these next couple of days no matter how realistic the proof may look.

Quote:

---

Originally Posted by SP33DFR34K

Oh I made them to prove a point. Very easy to make REALLY believable images. So don't believe everything you hear these next couple of days no matter how realistic the proof may look.

---

Point taken but in reality, it depends on the source.

Nobody is going to believe that DT or AT are faking screens, for example. FUDziller or Inq maybe:D

Which is pretty much common sense:up:

Quote:

---

Originally Posted by hawkeyefan

Point taken but in reality, it depends on the source.

Nobody is going to believe that DT or AT are faking screens, for example. FUDziller or Inq maybe:D

Which is pretty much common sense:up:

---

Of course, even as much as I am rooting AMD for their next gen processors, there is no way they can pull that kind of time with 2Ghz because from what I have experience, AMD generally doesn't do so well with SuperPi. So something that beats a 4Ghz Core 2 Duo plus the performance drop of using an AMD processor for Super Pi boggles the mind.

Common sense + Logic these next couple of days please :)

Quote:

---

Originally Posted by SOLDNER-MOFO64

Have faith mate......13s will be slow for these babies :)

---

Are you saying @ 2ghz the AMD k10 quad core will finish pi 1m in under 13 seconds? That would be amazing.

coincidently, If it does, Will pi become relevant as a benchmark again?

Quote:

---

Originally Posted by gallag

Are you saying @ 2ghz the AMD k10 quad core will finish pi 1m in under 13 seconds? That would be amazing.

coincidently, If it does, Will pi become relevant as a benchmark again?

---

All Benchmarks are relevant. There will never be a fair benchmark that can bench two different CPUs.

Quote:

---

Originally Posted by gallag

Are you saying @ 2ghz the AMD k10 quad core will finish pi 1m in under 13 seconds? That would be amazing.

coincidently, If it does, Will pi become relevant as a benchmark again?

---

From the benches I've been priveledged enough to see the results of (:))a 2ghz K10 will score around a 19sec 1M Pi time......at 3ghz approx 12secs.

:up:

Quote:

---

Originally Posted by SOLDNER-MOFO64

From the benches I've been priveledged enough to see the results of (:))a 2ghz K10 will score around a 19sec 1M Pi time......at 3ghz approx 12secs.

:up:

---

If that happens to be true then god bless AMD.

Quote:

---

Originally Posted by SP33DFR34K

Common sense + Logic these next couple of days please :)

---

QFT:up:

Quote:

---

Originally Posted by SOLDNER-MOFO64

From the benches I've been priveledged enough to see the results of (:))a 2ghz K10 will score around a 19sec 1M Pi time......at 3ghz approx 12secs.

:up:

---

Not often you get someone staking their reputation on the line like that. You better be right else you can expect no one believing a word from you in the future.

After ALL these threads about Barcie flowing around the XS, finally what is the REAL Super-Pi 1M?......39sec?.....32sec?......24sec?.....15sec?. .....13sec?.....What?.....Coz we've seen all of them and I'm wondering if I have to start saving for the Barcie platform OR to start opening my big mouth about AMD.....:p: :D

Quote:

---

Originally Posted by hipro5

After ALL these threads about Barcie flowing around the XS, finally what is the REAL Super-Pi 1M?......39sec?.....32sec?......24sec?.....15sec?. .....13sec?.....What?.....Coz we've seen all of them and I'm wondering if I have to start saving for the Barcie platform OR to start opening my big mouth about AMD.....:p: :D

---

4 days left my friend for the show.. :D

I hope AMD will do 13Sec @ 2000Mhz SPI, and you start BENCHEs with AMD CPU... :rofl:

Barcelona (2GHz) -> Pi 1M 13 secs....

Barcelona (3GHz) -> Pi 1M ?? secs....


Man, this is just to good to be true.... i hope it is, though....

Quote:

---

Originally Posted by Motiv

Not often you get someone staking their reputation on the line like that. You better be right else you can expect no one believing a word from you in the future.

---

I can appreciate that.......but what do I get when it turns out I'm perfectly correct? :D

Quote:

---

Originally Posted by hipro5

After ALL these threads about Barcie flowing around the XS, finally what is the REAL Super-Pi 1M?......39sec?.....32sec?......24sec?.....15sec?. .....13sec?.....What?.....Coz we've seen all of them and I'm wondering if I have to start saving for the Barcie platform OR to start opening my big mouth about AMD.....:p: :D

---

It'll be nice to see you HARD-CORE guys using AMD again.......shame really though.....I wanted to be THE MAN :rofl:

I wonder if KP will start to mess with K10.......surely he will huh? Or is he on INTEL's payroll? :yepp:

Quote:

---

Originally Posted by SOLDNER-MOFO64

I wonder if KP will start to mess with K10.......surely he will huh? Or is he on INTEL's payroll? :yepp:

---

k|ngp|n plays with what is faster.....;)

I, on the other hand, am an "Intel guy".... :p:

Quote:

---

Originally Posted by hipro5

k|ngp|n plays with what is faster.....;)

I, on the other hand, am an "Intel guy".... :p:

---

I'm sure INTEL are very glad you chose them Hipro.....I know AMD could use a few guys like you to brighten up their days ;):up:

Quote:

---

Originally Posted by Blasar

Barcelona (2GHz) -> Pi 1M 13 secs....

Barcelona (3GHz) -> Pi 1M ?? secs....


Man, this is just to good to be true.... i hope it is, though....

---

I highly doubt this. Because Super Pi...

• uses x87 code, which didn't get any improvement in K10
• has a dataset of several MB, while K10 only brings less of the fast L2 and just 2 MB of L3 (although nearly fully available for a single Super Pi task) to the table

These are just the probably most important reasons. My prediction: K10 won't be much faster than K8. But double digit percent faster is not unlikely thanks to other core improvements.

Anandtech's editor said that a newer barcelona stepping @ 2ghz does either 26 seconds in SuperPI
or 39-26=13 seconds. While the 13seconds @ 2ghz looks SF, the 26 seconds can be true.
i said "either" cause what he said is interpretable:)

4 more days to go:D

Quote:

---

Originally Posted by MR_SmartAss

I think the 39s is the most accurate, but it might be few secs faster. K10 does not have theoretical potential to perform like Core2 in SuperPi, but some people have too much faith in K10.

---

yes it is 100&#37; accurate for an older stepping wich the guy from coolaler's forum used:)
since then many things improved with new steppings, according to guys holding
these CPUs.

26 seconds:D

K10 cpu @ 3Ghz & DDR2 @ 1066+mhz 5,5,5,18 & AM2+ MOBO = <17s

K10 cpu @ 2Ghz & DDR2 @ 667mhz 5,5,5,15 & AM2 MOBO = 26

:D

Quote:

---

Originally Posted by SP33DFR34K

Of course, even as much as I am rooting AMD for their next gen processors, there is no way they can pull that kind of time with 2Ghz because from what I have experience, AMD generally doesn't do so well with SuperPi. So something that beats a 4Ghz Core 2 Duo plus the performance drop of using an AMD processor for Super Pi boggles the mind.

Common sense + Logic these next couple of days please :)

---

lol I 100% agree! This is where people get their panties all in a wad over inflated bench numbers only to be disappointed. When it launches we will know for the most part, barring bad reviews and or benches. That should be quickly sorted and in the end user hands we'll have good data to compare. IMO if AMD can approach Core2 4mb/quad performance clock for clock it's a win for them. NOT that it will win over the overclocking community, but 95% of sales is Dell/Best buy/HP type junk and they need to be more competitive rather than just nuking their ASP's of their cpu's. AMD's success doesn't rest on anyone in this forum, as much as you may think. Most of their failure can be chaulked up to horrible marketing, and business decisions(ati), not engineering. A64 was revolutionary in many respects when it hit. Now Core2, while being in many ways a decendent of a PIII was revolutionary in regards to performance when it hit. Will Barcelona/Phenom leapfrog Intel? I really doubt it. Again, if they close the performance gap that exists now clock for clock I think that will be enough for now. Overclocking I'm sure will stay on Intel's side, as I doubt these new chips will have the headroom to hit 3.6ghz on air that intel is getting, or more!
I hope to have some Barcelona/Phenom samples before long, but until I play with them and more info it known, it's all just speculation unless your in the "know" ;).

Guys the 13sec is FAKE. By some miracle of alien technology that AMD uses a 2Ghz AMD chip can pull 13 sec on SuperPi, everyone will start wearing their pink tutus in celebration.

Quote:

---

Originally Posted by SP33DFR34K

Guys the 13sec is FAKE. By some miracle of alien technology that AMD uses a 2Ghz AMD chip can pull 13 sec on SuperPi, everyone will start wearing their pink tutus in celebration.

---

I agree, and by some miracle if 13sec at 2ghz is true I'll wear my daughters pink ballerena outfit and dance around the neighborhood lol!

Quote:

---

Originally Posted by JNav89GT

I agree, and by some miracle if 13sec at 2ghz is true I'll wear my daughters pink ballerena outfit and dance around the neighborhood lol!

---

QFT

[Hopeisthelastthingtodie MODE ON]

you gonna make me take a plane to see that...:ROTF:

Start practicing some moves...

[Hopeisthelastthingtodie MODE OFF]

I have to admit that i dont believe in 13sec SPI for K10@2GHz, but i dont believe in Santa, either...:)

Quote:

---

Originally Posted by SOLDNER-MOFO64

From the benches I've been priveledged enough to see the results of (:))a 2ghz K10 will score around a 19sec 1M Pi time......at 3ghz approx 12secs.

:up:

---

Quote:

---

Originally Posted by SOLDNER-MOFO64

K10 cpu @ 3Ghz & DDR2 @ 1066+mhz 5,5,5,18 & AM2+ MOBO = <17s

K10 cpu @ 2Ghz & DDR2 @ 667mhz 5,5,5,15 & AM2 MOBO = 26

:D

---

????????????????????????????????????

only a few more days.

Im soo anctious to see these chips!

Quote:

---

Originally Posted by Gary Kelly

Depending upon the application, Barcelona clock for clock is equal to Clovertown or just a tad faster in some areas and in others it is 20&#37;+ (heavy emphasis on "plus" until Monday, ) faster. I have to say that final silicon is looking really nice at this point, especially in memory sensitive applications where this processor shines, but core speeds need to come up in a hurry to compete with Tigerton in general server applications.

The memory bandwidth numbers we noticed are a little surprising considering the results in certain memory intensive applications. The latencies are much improved over the last core stepping we looked at as is pure throughput which makes you think twice about the relationship between bandwidth and performance in general from this CPU.

My personal opinion is that if AMD had hit their original launch targets and speeds, the server market would be incredibly competitive at this point with Tigerton being a much needed response to Barcelona. As it stands, AMD is going to have an uphill battle at this time. But this is all subjective thinking from a guy who tried to compare an engine to a CPU, but maybe the results on Monday will clarify some of the observations I was trying to convey without breaking the rules.

---

Link
Another bit of info for the jigsaw puzzle

:)

This is getting out of hands, someone seen a real chip yet

Quote:

---

Originally Posted by bokis

This is getting out of hands, someone seen a real chip yet

---

Those who have one or more of the newer revisions (Dave Graham, Gary Key, et al) seem to like it. But NOBODY and I mean NOBODY is going to break NDA to satisfy the curiosity of a bunch of overclockers. We're just going to have to wait till Monday for a glimpse of some sort of reality.

A handful of those who don't have a chip spend 24 hours a day on here engaged in some of the wildest bunch of speculation I believe that I've ever witnessed in my entire life.

Both sides at this point...have gone completely insane...blinded by impatience. And patience is a virtue .:up:

One way or the other, a whole bunch of folks on XS are setting themselves up for a fat meal of crow next week.

Speculation = bad. Facts = good.

Heated arguments and personal attacks based on wild speculation = retarded.

I've got around a &#163;1000 saved up for these babies :D... so they better not disappoint!

Quote:

---

Originally Posted by hawkeyefan

Those who have one or more of the newer revisions (Dave Graham, Gary Key, et al) seem to like it. But NOBODY and I mean NOBODY is going to break NDA to satisfy the curiosity of a bunch of overclockers. We're just going to have to wait till Monday for a glimpse of some sort of reality.

---

That's assuming we get to see retail/desktop/client side chips for performance, if AMD only allows server benchmark reviews the debate will only get more heated.

Hopefully we will see not only the SPEC suit, but Black-Scholes, Apache and others, as well as some desktop applications....

This whole saga has been fun to watch....

Quote:

---

Originally Posted by JumpingJack

That's assuming we get to see retail/desktop/client side chips for performance, if AMD only allows server benchmark reviews the debate will only get more heated.

Hopefully we will see no only the SPEC suit, but Black-Scholes, Apache and others, as well as some desktop applications....

This whole saga has been fun to watch....

---

I know it, and I'm intentionally NOT getting my hopes up that we'll have some sort of definitive snapshot of K10 performance by 5 p.m. Monday:D

Now I'm hoping against hope that even if AMD doesn't put out a full suite of relevant benchies...at least the Dave Grahams and their ilk will be able to show us a few desktop benchies with NDA lifted.

Yeah, this thing has been a real trip the last few months, that's for sure. I think we have about half a dozen people accounting for thousands and thousands of posts over that time.:eek: :rofl:

I will say this much though...reading through all the crazy arguments...there is an incredible amount of stuff I've learned. Makes you appreciate how knowledgeable our forum members are...it's pretty incredible really. Some smart, smart dudes hanging around here. Even if they are insane fanboys one way or the other:up: :p:

Quote:

---

Originally Posted by hawkeyefan

I will say this much though...reading through all the crazy arguments...there is an incredible amount of stuff I've learned. Makes you appreciate how knowledgeable our forum members are...it's pretty incredible really. Some smart, smart dudes hanging around here. Even if they are insane fanboys one way or the other:up: :p:

---

Hence the reason I hang out in the forums at times :) ...

Jon Stokes made an amusing quip in his Tigerton launch article,

Quote:

---

I think we'll see that in the near-term, four-socket Barcelona systems will dominate four-socket Xeon systems in specfp_rate and in floating-point performance/watt. CPU-bound workloads will tell a different story, however, which is why I expect to see fanboys in both camps stinking up the Internet with specfp_rate versus spec_fp debates following the September 10 launch.

---

http://arstechnica.com/news.ars/post...-platform.html

Gave me a bit of a chuckle....

Jack

Quote:

---

Originally Posted by JumpingJack

Hence the reason I hang out in the forums at times :) ...

Jon Stokes made an amusing quip in his Tigerton launch article,

http://arstechnica.com/news.ars/post...-platform.html

Gave me a bit of a chuckle....

Jack

---

Right? Like the arguing will stop Monday:rofl:

As far as learning from these boards...I used to work for a bricklayer who would teach me some certain thing and then say "see, you learned something today. You had a reason to get out of bed this morning".

XS forums always give me a reason to get out of bed in the morning.

Oh well then we wait and see as usual, I am not a superPi freak, my main intrest is how well the preform in real life, fore instans in a htpc/mediapc
and offcourse the $$$$;)

Since when a single sPi result is enough to judge a whole CPU architecture?
Please grow up and w8 some serious real life benchies....

Quote:

---

Originally Posted by Fatal Error

Since when a single sPi result is enough to judge a whole CPU architecture?
Please grow up and w8 some serious real life benchies....

---

Exacly:clap:

Quote:

---

AMD Phenom to be available in small scale by October
From our friends in the industry, it is said that small quantities of Phenom processors should be available on sale as early as October.

There is no mentioning of the model and the price and exact availability date.
Looks like we will have to wait.

---

http://my.ocworkbench.com/bbs/showth...threadid=67137

End of October for Phenom loving :) (at least some lucky dudes will be able to play with it)

Quote:

---

Originally Posted by informal

http://my.ocworkbench.com/bbs/showth...threadid=67137

End of October for Phenom loving :) (at least some lucky dudes will be able to play with it)

---

great news if it pans out, thanks informal. Finally something from the guys in green BEFORE we expected it:D

Quote:

---

Originally Posted by SP33DFR34K

Of course, even as much as I am rooting AMD for their next gen processors, there is no way they can pull that kind of time with 2Ghz because from what I have experience, AMD generally doesn't do so well with SuperPi. So something that beats a 4Ghz Core 2 Duo plus the performance drop of using an AMD processor for Super Pi boggles the mind.

Common sense + Logic these next couple of days please :)

---

I have an interresting question i think, do you own any of those babies or it's full photoshopped ?

If it's not a full photoshop, do you overclocked one and decreased the frequency before taking the pic ? :rolleyes:

If it's fully photoshoped picture, so congratz, great work :D

Thx ;)

Quote:

---

Gary Key from AnandtechDepending upon the application, Barcelona clock for clock is equal to Clovertown or just a tad faster in some areas and in others it is 20&#37;+ (heavy emphasis on "plus" until Monday, ) faster. I have to say that final silicon is looking really nice at this point, especially in memory sensitive applications where this processor shines, but core speeds need to come up in a hurry to compete with Tigerton in general server applications.

The memory bandwidth numbers we noticed are a little surprising considering the results in certain memory intensive applications. The latencies are much improved over the last core stepping we looked at as is pure throughput which makes you think twice about the relationship between bandwidth and performance in general from this CPU.

My personal opinion is that if AMD had hit their original launch targets and speeds, the server market would be incredibly competitive at this point with Tigerton being a much needed response to Barcelona. As it stands, AMD is going to have an uphill battle at this time. But this is all subjective thinking from a guy who tried to compare an engine to a CPU, but maybe the results on Monday will clarify some of the observations I was trying to convey without breaking the rules.

---

Translation, good clock for clock performance compared to Clover, bandwidth numbers aren't astounding, still needs to ramp up speed to compete.

well, judging from what i've now experienced, i REALLY hate some of these pre-release bios' that only enable 1 physical CPU to be used...

(then again, it's fun seeing "Dual Core Opteron 1247 Core Count: 4" show up on the machine. :)

cheers,

dave

Haha,thanks for that input Dave :)

if you only knew the absolutely insane struggles i've had this evening.....

i've got 3 different boards (S2915, L1N64, S3992), 3 different beta BIOS (1.02C, 05.06, 2.00c), and 2 x FX-70, 2 x 2210, 4 x 2347, 2 x 2344 HE and i'm having to hand load XP Pro SP2 on the 2210 (solo) because of kernel panics on the CPUs and mainboards.

cheers,

dave

Quote:

---

Originally Posted by misteroadster

I have an interresting question i think, do you own any of those babies or it's full photoshopped ?

If it's not a full photoshop, do you overclocked one and decreased the frequency before taking the pic ? :rolleyes:

If it's fully photoshoped picture, so congratz, great work :D

Thx ;)

---

Full photoshoped. If i had these chips in hand, I wouldn't be sleeping :D

Quote:

---

Originally Posted by dave_graham

if you only knew the absolutely insane struggles i've had this evening.....

i've got 3 different boards (S2915, L1N64, S3992), 3 different beta BIOS (1.02C, 05.06, 2.00c), and 2 x FX-70, 2 x 2210, 4 x 2347, 2 x 2344 HE and i'm having to hand load XP Pro SP2 on the 2210 (solo) because of kernel panics on the CPUs and mainboards.

cheers,

dave

---

Damn,that is frustrating.Any hope the BIOS support will be better after the official launch?

Thanks for sharing stuff here:up:

so, here's the happy news....managed to load out XP Pro with the 2210s, swapped the 2347 in and....guess what i'm staring at now?

yup....that damn XP loading screen....

yippee..

yes, BIOS support should be better. word on the street is the S2915 is getting a board respin so....i'll definitely need to check one of those out.

EDIT: 3 dead boards so far.....from a variety of ailments. *sniff, sniff*
dave

Quote:

---

Originally Posted by dave_graham

so, here's the happy news....managed to load out XP Pro with the 2210s, swapped the 2347 in and....guess what i'm staring at now?

yup....that damn XP loading screen....

yippee..

yes, BIOS support should be better. word on the street is the S2915 is getting a board respin so....i'll definitely need to check one of those out.

EDIT: 3 dead boards so far.....from a variety of ailments. *sniff, sniff*
dave

---

Da-am! After the NDA, please tell us what killed them?

They are not "dead"...The boards aren't working the way they should.That's all.

Quote:

---

Originally Posted by dave_graham

so, here's the happy news....managed to load out XP Pro with the 2210s, swapped the 2347 in and....guess what i'm staring at now?

yup....that damn XP loading screen....

yippee..

---

Try 2k3 ;)

By now, I would guess that everybody is holding their breaths and waiting for info from the ones well known and that clearly stated to habe those beauties on hands...

(and also that the ones they have on hands are different from the ones coolalers have!)

Hopefully, the NDA will not also kill everybody asfixiated cause of holding their breaths for too long... ;)

Quote:

---

Originally Posted by dave_graham

if you only knew the absolutely insane struggles i've had this evening.....

---

Sounds like exactly the kind of shape you'd want to launch your new server processor in ... NOT. :clap:

Quote:

---

Originally Posted by Ubermann

13s at Spi at 2ghz..yeah right =)
Some people need a reality check.

---

Whatever it would be, 2secs at 10Mhz or 2hours at 3Ghz, we want to see more realistic benchmarks. Superpi is nowhere near that. 3Ghz Conroe could do about 14secs, 3.5Ghz Windsor about 23secs. In most benchmarks they are about to equal. I know that there are few people who care about superpi, but lets be serious, until REAL DESKTOP PHENOM release all these faked screenshots are just waste of time.

"who plays 3D MARK ?!?!"
"SuperPi is useless"
"Sandra is useless"

Are you sure you are on the right forum when saying this kind of stuff ?
This site is 99.9 &#37; about benching and overclocking with the above apps.
Do you see many guys here benching games or other non bench apps?

You need to think about that..or maybe consider if there is any other forum that has a info that are more towards your requested games/apps or whatever.

(Not to you karparz but to many others)

Quote:

---

Originally Posted by Ubermann

"who plays 3D MARK ?!?!"
"SuperPi is useless"
"Sandra is useless"

Are you sure you are on the right forum when saying this kind of stuff ?
This site is 99.9 % about benching and overclocking with the above apps.
Do you see many guys here benching games or other non bench apps?

You need to think about that..or maybe consider if there is any other forum that has a info that are more towards your requested games/apps or whatever.

(Not to you karparz but to many others)

---

For me, I want to see the Folding@Home SMP performance :D. If it does better than the current Intel Quad Cores, I'm in.

Quote:

---

Originally Posted by SP33DFR34K

For me, I want to see the Folding@Home SMP performance :D. If it does better than the current Intel Quad Cores, I'm in.

---

Thats a big part of this site also so thats not a problem.
But you know what i mean.

Its like im going to a dog forum and say i like cats better.

These ones looks nice: http://forums.vr-zone.com/showthread.php?t=184594

That means the Barcelona should destroy the Clovertown in SuperPi (since it measures FP).

I heard the next Corvette Z06 was going to be powered by Dark Matter.

Confirm/deny?

Quote:

---

Originally Posted by AgentGOD

That means the Barcelona should destroy the Clovertown in SuperPi (since it measures FP).

---

I believe in SSE and FP applications it will be an outstanding CPU!
this means graphics applications and many of the applications
that really need more power.

but well see:)

Have I understood something wrong, or is it really IBM's dark matter? :D

http://forums.2cpu.com/showthread.php?t=85815

Quote:

---

Originally Posted by informal

They are not "dead"...The boards aren't working the way they should.That's all.

---

gee...ya think i don't know what boards do?

lesse.

S2915 A: never was verified to work before i got it; definitely does work with a single 2210, spec. memory.

S2915 B: see above.

S2915-L C: worked fine with .95 bios (it's a dev. board.) flashed to 1.02C beta and no workie. Needs a new BIOS chip (back to .95 since it's my core dev. board

3992 A: flashed Beta bios. worked but core panic. received updated production BIOS from Tyan AND AMD on Friday night. Testing today.

L1N64 : loops processor and doesn't work with 2347. haven't had the time to go back and test with my fx-70/2210

trust me, i know more about these boards than most.

cheers,

dave

Quote:

---

Originally Posted by MR_SmartAss

There are some issues with the nVidia chipsets, but I think that a BIOS update will fix them.

---

actually, its appearing more and more that it might be a fundemental issue with the board design COUPLED with the nVidia chipset.

cheers,

dave

Quote:

---

Originally Posted by Proesterchen

Sounds like exactly the kind of shape you'd want to launch your new server processor in ... NOT. :clap:

---

actually, it was half expected due to early warning notes from Tyan and AMD.

i chose to press on and that's where I found myself. :)

dave

Benchs from IBM, original links:

ftp://ftp.software.ibm.com/eserver/b...cpu_091007.pdf
ftp://ftp.software.ibm.com/eserver/b...yna_091007.pdf

Anand released it! http://www.anandtech.com/cpuchipsets...oc.aspx?i=3092

WHAT THE F????

15&#37; faster than k8? clock for clock....are they joking?

Will a single chip work with a 590 SLi AM2 board do you think? Its just that hearing you say that there are problems with nVidia chipsets worrys me a little bit, although I realize you are primarily testing server boards.

Quote:

---

Originally Posted by n91htmare

WHAT THE F????

15% faster than k8? clock for clock....are they joking?

---

Is that about the same as a C2D? - so if it REALLY overclocks well,... :/ it'll be ok?

Yes, it is (theoretically) about the same as C2D in integer, though looking to be better in FP. Anandtech's review might be a preview, but it's kinda a half-assed way to do things, if not a little inventive. What we need are AM2 compatible processors out in the wild that can be more easily benched and compared to a Core 2 Processor in terms of performance, graphics, etc.

15&#37; puts it close to C2D, not to perryn... but from 667 ECC to 800 non-ECC there will be differences, and I would not guess how large they will be.

For my taste Barcellona is already showing what it came for: servers. :) And I WANT ONE!



Ok, not a Barcellona, rather more a Shangai or, better, a Montreal. :D

Anyway: first benchies are now out, official, from Anand and IBM, and Anand showed that there already are 2.5GHz beaties around... :D

If it costs AMD more to produce it AND they don't have the brand name to sell tonnes, then they need to undercut intel (still) therefore it's really just going to keep them scraping by :/ - unless it clocks to 4.0ghz easily.

They are stuffed.

released from anand

there is no matter (usually) in OC when we talk about server processors. The MBs doesn't cooperate, and the processor also do not help much. So, I would not expect to see OCs on those babies soon.

Unfortunatelly, of course... :D

Quote:

---

Originally Posted by AbRASiON

If it costs AMD more to produce it AND they don't have the brand name to sell tonnes, then they need to undercut Intel (still) therefore it's really just going to keep them scraping by :/ - unless it clocks to 4.0ghz easily.

They are stuffed.

---

Not at all. If the server performance is what it is looking to be, AMD has the possibility of gaining more market share. Quite honestly, this is the sector where money is made. The consumer market obviously means a decent amount of cash, but the net profit per chip is drastically less. Oh, and overclocking means next to nothing for 99% of the consumer market, much less the business sector... :up:

Quote:

---

Originally Posted by xVeinx

Not at all. If the server performance is what it is looking to be, AMD has the possibility of gaining more market share. Quite honestly, this is the sector where money is made. The consumer market obviously means a decent amount of cash, but the net profit per chip is drastically less. Oh, and overclocking means next to nothing for 99% of the consumer market, much less the business sector... :up:

---

gee....i've only stated THAT about 1000 times. :)

you'd think they'd listen...

dave

Quote:

---

Originally Posted by xVeinx

Not at all. If the server performance is what it is looking to be, AMD has the possibility of gaining more market share. Quite honestly, this is the sector where money is made. The consumer market obviously means a decent amount of cash, but the net profit per chip is drastically less. Oh, and overclocking means next to nothing for 99% of the consumer market, much less the business sector... :up:

---

Rubbish, overclocking gives an indication of the head room of the CPU.
If it's barely equivelant to the performance of a C2D chip - the ONLY WAY AMD will sell them is to cost them lower than intel or to hardcore enthusiasts.
This is how it works, this is how it has worked forever between intel and AMD

Quote:

---

Originally Posted by AbRASiON

Rubbish, overclocking gives an indication of the head room of the CPU.
If it's barely equivelant to the performance of a C2D chip - the ONLY WAY AMD will sell them is to cost them lower than intel or to hardcore enthusiasts.
This is how it works, this is how it has worked forever between intel and AMD

---

fie on your overclocking!!!

it actually now has less to do with headroom than anything. for example, how will you manage the split plane power delivery since the sources are not synchronous anymore?

hmmmm?

guaranteed that not a lot of big OEMs are going to spend the time oc'ing a chip on water, air, nitro, etc. to see "headroom." they're going to want X processor to run at the rated speed, 24x7x365 without error. overclocking, while somewhat useful in that regard, is a needless waste of time.

dave

Quote:

---

Originally Posted by dave_graham

fie on your overclocking!!!

it actually now has less to do with headroom than anything. for example, how will you manage the split plane power delivery since the sources are not synchronous anymore?

hmmmm?

guaranteed that not a lot of big OEMs are going to spend the time oc'ing a chip on water, air, nitro, etc. to see "headroom." they're going to want X processor to run at the rated speed, 24x7x365 without error. overclocking, while somewhat useful in that regard, is a needless waste of time.

dave

---

From a DESKTOP PERSPECTIVE, if the AMD CPU is the same speed as the intel C2D (finally!) at the same clock speed, the only way their CPU can be awesome is

A, it clocks much higher than intels offering
B, it costs a lot less.

This is how it's been for a while.

So is this an actual hard-launch or just the end of the NDA? To put it more bluntly, where's the freakin benchmarks???

Get the damn chips to someone other than Anandtech to test the heck out of em! Heck, I'll volunteer if needs be :p:

I think Dave has already all benchies he wants, and would be counting seconds to publish tit somewhere else and put the link here. :D

Quote:

---

Originally Posted by dave_graham

fie on your overclocking!!!

it actually now has less to do with headroom than anything. for example, how will you manage the split plane power delivery since the sources are not synchronous anymore?

hmmmm?

guaranteed that not a lot of big OEMs are going to spend the time oc'ing a chip on water, air, nitro, etc. to see "headroom." they're going to want X processor to run at the rated speed, 24x7x365 without error. overclocking, while somewhat useful in that regard, is a needless waste of time.

dave

---

All points are very true, but you're waisting you time. 95&#37; of your audience here has never thought about needing a server and more importantly thinks that their Overclocking world is all that's out there. No offense to anyone, I love to overclock, but what we do means very little to the real world. Anyway, I'm happy for AMD I'm sure they'll be able to keep thing interesting with their new cpu.