AMD Prices "Barcelona"

[dave_graham]

he's speaking about the coolaler results, which prolly *were valid* at a certain time, but retail barc will be higher, with a more mature platform/stepping - but by how much we don't know (that's the point of all the speculation).

coolaler was using early B0 or B1 results with a STREAM errata that impacted the benchmarks significantly.

there are at least 2 more revisions coming.

dave

[Lightman]

i've got two of them.

gotta love it

dave

Thanks for bringing reliable news Dave!

Are you gonna bench them to death?

Seriously thou what mobo you're using with them?
Can you check them on QuadFX mobo if you have one? Please!

[Shintai]

Correct me if I'm wrong, I remember hearing a while back that Intel rates its processor TDP differently than AMD. Intel supposedly goes based on average TDP, while AMD goes with max TDP. This could explain why AMD may have a higher TDP than Intel.

Wrong, both are max TDPs. You cant find a single Core/Core 2 CPU over its TDP. They are usually a good bit below.

but when AMD or Intel is pressed hard. They seem to be close or go abit over. P4 and late K8s as example.

but also remember AMD and Intel only have a few categories of TDPs. So a CPU could just barely be over a lower TDP, or maybe just at TDP and get a TDP rating 50% higher. Say a 81W CPU getting 120W TDP.

[BrowncoatGR]

Wrong, both are max TDPs. You cant find a single Core/Core 2 CPU over its TDP. They are usually a good bit below.

but when AMD or Intel is pressed hard. They seem to be close or go abit over. P4 and late K8s as example.

but also remember AMD and Intel only have a few categories of TDPs. So a CPU could just barely be over a lower TDP, or maybe just at TDP and get a TDP rating 50% higher. Say a 81W CPU getting 120W TDP.

Thats not exactly true. AMDs TDP is indeed supposed to be the max obtainable, while Intels is the max under normal use. Intels TDPs seem to be chosen quite well though. OTH AMDs CPUs also include the northbridge part so TDPs can not be directly compared.

[hawkeyefan]

but also remember AMD and Intel only have a few categories of TDPs. So a CPU could just barely be over a lower TDP, or maybe just at TDP and get a TDP rating 50% higher. Say a 81W CPU getting 120W TDP.

Excellent point.

[Shintai]

Thats not exactly true. AMDs TDP is indeed supposed to be the max obtainable, while Intels is the max under normal use. Intels TDPs seem to be chosen quite well though. OTH AMDs CPUs also include the northbridge part so TDPs can not be directly compared.

Show me a P-M/Core/Core 2 that is even near its max TDP even with TAT.

Its BS and bollocks. The only thing you can show is P4s, and in the same way I can show you late highend K8s doing the same (Specially 4x4 CPUs). So its an old rumour story at best in terms of lack of better understanding.

[Proesterchen]

Wrong, both are max TDPs.

Actually, AMD and Intel are both now quoting max continuous power consumption over a thermally significant period using commercially available software (or something to this effect), which is below either's max output.

You cant find a single Core/Core 2 CPU over its TDP. They are usually a good bit below.

Indeed.

Thats not exactly true. AMDs TDP is indeed supposed to be the max obtainable, while Intels is the max under normal use.

^ See above. AMD hasn't been publishing TDP max for years now.

[kl0012]

Thats not exactly true. AMDs TDP is indeed supposed to be the max obtainable, while Intels is the max under normal use. Intels TDPs seem to be chosen quite well though. OTH AMDs CPUs also include the northbridge part so TDPs can not be directly compared.

You missing that the same TDP (65W) is listed for all dual core desktop parts - from 1.6 GHz to 3GHz. It can't be just "max under normal use" because Pentium E2140 consumes far less power then E6850 even under "extreme use".

[dave_graham]

That guy from coolaler was benching B1. It is a good news to hear that new revisions will arrive. What do you think, how much can AMD improve performance in these 2 or more revisions?

one rev is for the higher speed parts and one is launch.

all that I can legally say is that the STREAM performance impact (most readily shown by coolater's test) has been mitigated.

cheers,

dave

[Dimitriman]

Prices are good
Lets hope the performance is also =]

[Dimitriman]

one rev is for the higher speed parts and one is launch.

all that I can legally say is that the STREAM performance impact (most readily shown by coolater's test) has been mitigated.

cheers,

dave

you put hope into many people hehe

lets hope that new revision for higher clocks comes with desktop launch

[Aussie FX]

Would you like to explain more about the STREAM errata and its performance impact. Thanks

LOL, read what he said.

Basically those benches from coolalers site are bs. It is nothing near the actual performance of retail chips.

[savantu]

LOL, read what he said.

Basically those benches from coolalers site are bs. It is nothing near the actual performance of retail chips.

BS.

Bx parts work fine ; A0 had L3 disabled and HT links running at crazy low frequency/not at all.

[informal]

BS.

Bx parts work fine ; A0 had L3 disabled and HT links running at crazy low frequency/not at all.

Reread Dave's posts again,than come back and write that again.

[Aussie FX]

Reread Dave's posts again,than come back and write that again.

savantu is correct. (I think)

There were es chips that had all sorts of things disabled on them.
Some had disabled l3 others had HT downclocked etc.

That's why you can't really take any notice of early revision chips because you wouldn't know what was disabled, unless of course AMD informed you.
That's how I understand it anyway.

[Aussie FX]

I did, so?

Sorry what I meant was Dave said he couldn't say anymore.

NDA??

[informal]

savantu is correct. (I think)

There were es chips that had all sorts of things disabled on them.
Some had disabled l3 others had HT downclocked etc.

That's why you can't really take any notice of early revision chips because you wouldn't know what was disabled, unless of course AMD informed you.
That's how I understand it anyway.

No he is not correct.He said "Bx parts work fine" .BUT Mr. Dave Graham said in no uncertain terms,that Bx wasn't 100% bug free(not only chips but motherboards that weren't operating with the appropriate 100% working BIOS).

[Jacky]

No he is not correct.He said "Bx parts work fine" .BUT Mr. Dave Graham said in no uncertain terms,that Bx wasn't 100% bug free(not only chips but motherboards that weren't operating with the appropriate 100% working BIOS).

complex ICs like cpus/chipsets never are 100% bug free (errata), but it always depends on how serious/performance limiting those bugs are.

and we still don't know if the coolaler cpus are not from the september-launch stepping

[informal]

and we still don't know if the coolaler cpus are not from the september-launch stepping

Yes we do.B1 is not the launch stepping....

[Jacky]

Yes we do.B1 is not the launch stepping....

are you sure? actually it's not that important, because we'll find out really, really soon

[PetNorth]

are you sure? actually it's not that important, because we'll find out really, really soon

http://www.xtremesystems.org/forums/...&postcount=510

http://www.xtremesystems.org/forums/...&postcount=513

[informal]

are you sure? actually it's not that important, because we'll find out really, really soon

Yes i am sure.Look at the links PetNorth provided.

[BrowncoatGR]

Actually, AMD and Intel are both now quoting max continuous power consumption over a thermally significant period using commercially available software (or something to this effect), which is below either's max output.

http://www.xtremesystems.org/forums/...36&postcount=1
**Edit** At least for Server/Workstation CPUs AMD uses max TDP. If they dont use it for consumer cpus though that would explain why Opterons have higher TDP than equivalent desktop parts. Look at the bottom slide V_rr posted.

[savantu]

http://www.xtremesystems.org/forums/...&postcount=510

http://www.xtremesystems.org/forums/...&postcount=513

Completely irrelevant ; it doesn't matter if B1 is launch stepping or not , it matters how it performs.Based on everything available , B1 performs normally but has problems scaling in frequency.

[alayashu]

Completely irrelevant ; it doesn't matter if B1 is launch stepping or not , it matters how it performs.Based on everything available , B1 performs normally but has problems scaling in frequency.

Based on everything available, until B2 all previous steppings had serious problems
affecting performance considerably. Both chip and bios problems.