AMD Ships First "Interlagos" CPUs based on Bulldozer

[zanzabar]

amd has a share of about 30% of the cores sold (and intel says anything with ecc is a server so its not really a good representation, like the graphics sales numbers), they are the only pick for VM farms and the BD killed the last to quarters since there is no reason to buy an upgrade until BD is out for a vm farm so alot of companies and data centers were waiting. amd also was supposed to have the entire 1st few batches already sold.

on the speed comment, most chips are stable around 2-2.5ghz now with the 1st batch of silicon, but when u go up to 3ghz+ then u run into problem with to much or not enough leakage and the way the cashe and fixed clock parts interact with the cores. so when your part is based for mid TDP servers and your side market is the desktop enthusiast that wants something clocked higher and not what the chip was primarily designed to do, your going to make sure that every thing is nice for the clock ranges it need to be 100% then u work a second rev into the more clock and voltage demanding market.

[Glow9]

hey, what about april 17th? Do you guys think amd will launch bd before that?

lol...

[Hornet331]

amd has a share of about 30% of the cores sold (and intel says anything with ecc is a server so its not really a good representation, like the graphics sales numbers), they are the only pick for VM farms and the BD killed the last to quarters since there is no reason to buy an upgrade until BD is out for a vm farm so alot of companies and data centers were waiting. amd also was supposed to have the entire 1st few batches already sold.

But you neither amd nor intel get payed by core only per cpu.

[Shocker003]

Well this is a copy of the e-mail i got from AMD and i guess they might announce BD on this day
I would have loved to but i live in Germany and annual leave has been planned out

Kay, hanging out in San Francisco the week of September 12th? Not finding anything interesting?
AMD to the rescue. We'll be making an historic announcement, and want you to be a part of it.

AMD invites you to join us for an entertaining evening on the beautiful Yerba Buena Terrace at the St. Regis San Francisco . Spend the evening exploring the latest AMD technology, mingling with AMD executives and technology partners, all while enjoying tails(male-chicken-tail) and hors d'oeuvres. Be sure to arrive before 7:00pm to hear our big news first hand.
When: Tuesday, September 13, 2011
Where: St. Regis Hotel, Yerba Buena Terrace, San Francisco
Time: 6pm - 9 pm PDT
RVSP: by September 9, 2011 at fusionzone.eventbrite.com (password:)
Contact Information:
Heather J Lennon
Sr. Manager Public Relations, AMD
Heather.Lennon@amd.com



Advanced Micro Devices, Inc., One AMD Place, P.O. Box 3453, Sunnyvale, California 94088-3453 USA
AMD Developer Central | RSS | Copyright 2011 | Privacy | Trademark Info

[EniGmA1987]

^^^^ The event is under Fusion though, and Zambezi isnt a fusion processor. That would be more along the lines of Llano. So maybe it has to do with something new like the 2nd gen fusion AMD has had on the books or maybe something else. Id be going but it is on a Tuesday and I cant drive 5 hours to go then drive 5 hours back in one day. Its just too much driving for a small event.

[BeepBeep2]

^^^^ The event is under Fusion though, and Zambezi isnt a fusion processor. That would be more along the lines of Llano. So maybe it has to do with something new like the 2nd gen fusion AMD has had on the books or maybe something else. Id be going but it is on a Tuesday and I cant drive 5 hours to go then drive 5 hours back in one day. Its just too much driving for a small event.

Nothing except the website to RSVP at says anything about Fusion.

[[XC] Synthetickiller]

Is there any info on the new opertons speed / prices?

I have a G34 8core and would love an excuse to double the cores if the clock speed is much higher (3ghz please!)

[qcmadness]

Is there any info on the new opertons speed / prices?

I have a G34 8core and would love an excuse to double the cores if the clock speed is much higher (3ghz please!)

16-core products will not reach 3GHz in 2011.

[informal]

16-core products will not reach 3GHz in 2011.

Top bin Interlagos that will launch in Q4 (2.3Ghz) will reach 2.8Ghz on all integer cores and 3.3Ghz on 8 active integer cores. Turbo won't kick in when all FPUs(or FlexFPs) are being loaded though. I expect that in Q1 AMD releases 2.5/2.6Ghz 16C Interlagos models.

[[XC] Synthetickiller]

Top bin Interlagos that will launch in Q4 (2.3Ghz) will reach 2.8Ghz on all integer cores and 3.3Ghz on 8 active integer cores. Turbo won't kick in when all FPUs(or FlexFPs) are being loaded though. I expect that in Q1 AMD releases 2.5/2.6Ghz 16C Interlagos models.

Thanks!

[Mech0z]

Shouldnt NDA for server parts be lifted when the products are shipped? Btw is it possible to buy any cheap G34 boards in retail? I have a hard time finding any for my private server.

[bamtan2]

^^^^ The event is under Fusion though, and Zambezi isnt a fusion processor. That would be more along the lines of Llano. So maybe it has to do with something new like the 2nd gen fusion AMD has had on the books or maybe something else. Id be going but it is on a Tuesday and I cant drive 5 hours to go then drive 5 hours back in one day. Its just too much driving for a small event.

the word fusion has no meaning. the ceo just said fusion means their whole company. it means nothing.

[Dimitriman]

Sisoft Benchmarks of "Interlagos" Opteron 6282 SE (Which I do not know what frequency it should have but apparently it is 2.5Ghz base clock)

http://www.sisoftware.eu/rank2011d/s...fb9caf7c7&l=en

I have no idea if they are good or what, would anyone care to compare with Intels?

Summary:
* Processor Arithmetic Benchmark: 231.971 GOPS
* Processor Multi-Media Benchmark: 585.983 Mpix/s
* Multi-Core Efficiency Benchmark: 21.773 GB/s
* Cryptography Benchmark: 4661 GB/s

Thanks.

[informal]

I have tried to compare it to 12 2.5Ghz MC system here.

Nice find. At what clock speed is that thing running? If AMD follows MC naming scheme it should be 2.5Ghz. Sisoft detects it as 2.0Ghz though. Anyway,for comparison purpose:
2x 12C Opteron @ 2.5Ghz
2x 12C Opteron @ 2.3Ghz

If this sample is running at 2.5Ghz then FP performance is around 35% better per core and per clock than MC. Kinda conflicting with what we see from Zambezi and Cinebench/wprime,ain't it?
Integer throughput is a bit lower per clock though. This is if we presume it runs at the same clock as MC. If it runs at 2Ghz then it is 7.4% faster per core in this trhoughput workload(and if 80% sclaing thing is true for integer,then each core is 1.074x1.2~=1.3x or 30% faster than MC core). Cryptography is much lower than what that ES 16C Interlagos got in July,around 50% lower. Something is affecting the scores.It could be that every sample has been "marked" in some way,so AMD knows exactly who leaked what .

[Dimitriman]

So... Are you saying IPC decreases??

[informal]

Well yes and no . It depends at what clock that damn 6282SE thing runs at. If it's 2Ghz base clock then IPC increases in both int and fp(int slightly ,by 7.5% and FP tremendously,by 66% !). If it runs at 2.5Ghz base clock then int decreases per clock slightly(at least in this one benchmark,by 7%) while FP still increases a lot,by 30+% per clock. Since I don't think 66% per clock increase is probable in fp without (FMA) recompile then I vote for 2Ghz base clock on this part,even though it goes against the Interlagos naming scheme(6276 is 2.3Ghz base clock part).

[Dimitriman]

Maybe they purposefully ran it at 2.0Ghz, also a possibilty (perhaps due to NDA?)

Hope so.

[informal]

Yes,that could be a possibility,especially now when Interlagos supports power cap feature. If you limit the chip in BIOS to say 75W TDP it will always run at the clock speed that is not putting it over this limit in full load (or full Turbo mode). Pretty nice feature.

[Dimitriman]

Yep agree, I was thinking of that specifically as well.

But in any way if 32 BD cores = 48 MC cores in most tests then its not bad at all.

[informal]

Yes I agree and it bodes well with previous statements of 50% more throughput. It's even more in FP workloads (and a bit less in integer ones so average is ~50%).

[informal]

Ok a small update on sisoft numbers for Opterons. I'm pretty sure now about following things:
1)Both Opteron sisoft results are real. Some features are turned off though.
2) 6220 part has a correct SIMD throughput,while 6282SE has somewhat inflated number (memory configuration is maybe specced higher there).
3) Both 6282SE and 6220 servers had Turbo off in integer tests. First ran at 2.5Ghz(2x 16C) ,second at 3Ghz(2x8C). Multimedia test uses AVX and gives 11% better score than legacy SSE.
4) The Opterons that will launch very soon will have roughly(2P top bin next gen vs 2P top bin previous gen) : 28-30% higher spec_int_rate score and 33-35% higher spec_fp_rate score. This is 2.3Ghz(2.8Ghz Turbo on all cores for integer workloads) Interlagos Vs 2.5Ghz 12C Magny Cours. In IPC numbers this is ~7-8% higher integer IPC and 8-10% higher floating point/SIMD IPC -all in non recompiled workloads. AVX 256/128brings ~10% more in floating point and FMA brings up to 2x more over AVX,but this is "what if case" and not the norm(we have to wait for applications to be written with FMA in mind). I guess XOP will bring similar speedups to AVX ,10% or maybe more,in integer recompiled workloads.
5) Zambezi's Sisoft results that were leaked are not correct. I don't know whether the Turbo was on or off in that test,but even if it was off the results are ~17-20% lower in integer part than what the opterons show. FP part is more or less correct since Opterons score the same per core and clock, but the test was run on ES platform with 1333Mhz DDR3 memory and unknown BIOS settings. Even if we take the legacy SSE score of 132Mpix/s,which was scored at fix clock of 2.8Ghz (100% sure about this) and correct it for launch clock of FX8150 part we arrive at 170Mpix/s.This is 54% more than 1100T.If we take best for Zambezi then it will be AVX. Now it is 71% faster than 1100T (stock versus stock => 189Mpix/s Vs 110Mpix/s).
Integer throughput is around 35% higher on FX8150 versus 1100T (stock+Turbo Versus stock=>88Gops Vs 65Gops).

I suppose numbers can be higher for desktop version,by about 5% ,compared to the ones I posted in point 5). As for Opterons,I'm 99% sure this will be the speed up that SPEC benchmarks will show. Oh and STREAM(memory BW) will be around 50% faster on Interlagos ,but this is already known.

PS And yes,this means Zambezi shouldn't score lower than Thuban 1100T in Cinebench... At least not according to above. But who knows,anything is possible.

[Dimitriman]

Ok a small update on sisoft numbers for Opterons. I'm pretty sure now about following things:
snip.

I hope you are right because I am already preparing my apologies to terrace who's predictions will all come true if Zambezi comes in October with barely faster than 1100T and with lower IPC...

Still I am hopeful this will not happen.