AMD ''Interlagos'' Bulldozer Benchmarks Leaked

[jimmysmitty]

http://www.tomshardware.com/news/Bulldozer-Interlagos-Phoronix-Test-Suite-Benchmark-16-core,12445.html

Interesting but I will lay some up front stuff to avoid stupid flaming and /or fighting:

1. This is 32 BD cores compared to 4 SB and 6 Nehalem cores. So its not ANY indication as to how well Zambezi (4-8 core BD units) will perform against current Intel stuff.

2. The clock speed is low but thats normal for AMD when they intro more cores into the server market. They always start off with low and move up higher where Intel replicates its desktop side in clock speed. Does not indicate that BD wont be able to clock higher.

Overall, i think its a pure BS benchmark. No idea why they would even think of comparing 4 cores 8 threads to 32 cores but who am I to disagree with them?

 

[ghnader hsmithot]

why does the release date keep on changing..You piece of crap AMD..We need BullDozer now!

 

[majin ssj eric]

I wish AMD would really bring it this next generation. I'm an Intel guy but they need AMD to push them....really

 

[jimmysmitty]

why does the release date keep on changing..You piece of crap AMD..We need BullDozer now!

Remember these are specifically the server CPUs, which is why its two 16 core Interlagos chips.

Desktop will have up to 8 cores codenamed Zambezi. It is supposed to be released in June of this year but no hard date has been given, so its still not a 100%.

I would like to see how Interlagos performs against Xeons, specifically 2 Nehalem EX CPUs vs two of these. With that we could see if a 4 core i7 could compete with a 8 core BD unit.

 

[ElMoIsEviL]

Anytime anyone posts a thread about Bulldozer... a Mod here (who is an AMD fan) rains on our parade and locks it (even if there is no arguing) making the claim that we "enthusiasts" cannot "speculate/be enthused" about upcoming products. I'm almost tempted to report him to his superiors.

In other words... this thread is likely to get locked.

 

[jimmysmitty]

This is legit though. Its even on THG itself.

Lets hope not and hope it doesn't fall into another back and forth crap fest.

 

[loneninja]

This data is meaningless to me, 32 extremely slow cores compared to 4-6 really fast ones, makes so much sense. lol

I'll get excited when it's at least server vs server or desktop vs desktop.

 

[Hellboy]

oh come on 64gb of ram and a dual socket board...

this is a bull s**t test

was there a dual sandybridge board with the same 64gb ram

i doubt it...

this is pure marketting crap.

 

[gamerk316]

Where psycho lol not 4 beats 8 its 4 beats 32 looool

In a situation where you can not thread well, 4 fast cores will almost always beat 32 slow cores, especially when the 4 core solution has a better IPC anyways...

The fact AMD didn't win across the board is a bit troubling, frankly, since PC software is much harder to thread then server software [which has been optimized for multiple CPU's for decades].

 

[Onus]

Unfortunately, performance information (good or bad, promising or not) is pretty meaningless without price information to go with it. Oh well, the wait goes on.

Edit: the fact that AMD didn't "win across the board" comes as no surprise, even though they won one test by a substantial margin. Their architecture looks sufficiently different from the tried-and-true that I rather suspect we're going to see the intended use make a much bigger difference when deciding which CPU to choose. It is possible that AMD will own the blade-server market with their latest generation (it looks like you can run a LOT of VMs with Bulldozer; can you say "cloud?"), while yielding even the low-margin desktop. This isn't what I want to see (I think "cloud" computing is generally a really bad idea for the mainstream), but may be what AMD wants. We'll see.

 

[Wisecracker]

Anytime anyone posts a thread about Bulldozer... a Mod here (who is an AMD fan) rains on our parade and locks it (even if there is no arguing) making the claim that we "enthusiasts" cannot "speculate/be enthused" about upcoming products. I'm almost tempted to report him to his superiors.

In other words... this thread is likely to get locked.

I suspect the thread will not get locked if folks stay on topic and cut the BS.

why does the release date keep on changing..You piece of crap AMD..We need BullDozer now!

This would be BS.

 

[Hellboy]

jf-amd we are open to comments here

 

[ghnader hsmithot]

i just want more competition in the market..it really sucks just having to see Sandy Bridge all the time..I want more choices!

 

[fazers_on_stun]

Hmm, these "leaked" benchies have been all over the board. I suspect this one is also bogus. Why didn't they release the other results if they actually benched those as well?

 

[jimmysmitty]

Unfortunately, performance information (good or bad, promising or not) is pretty meaningless without price information to go with it. Oh well, the wait goes on.

Edit: the fact that AMD didn't "win across the board" comes as no surprise, even though they won one test by a substantial margin. Their architecture looks sufficiently different from the tried-and-true that I rather suspect we're going to see the intended use make a much bigger difference when deciding which CPU to choose. It is possible that AMD will own the blade-server market with their latest generation (it looks like you can run a LOT of VMs with Bulldozer; can you say "cloud?"), while yielding even the low-margin desktop. This isn't what I want to see (I think "cloud" computing is generally a really bad idea for the mainstream), but may be what AMD wants. We'll see.

Well Intel does have "Cloud" specific CPUs. It was the 48 core they developed and sent out to reserchers that was based on Terascale:

http://www.pcworld.com/businesscenter/article/183653/intel_48core_singlechip_cloud_computer_improves_power_efficiency.html

http://www.tomshardware.com/news/48-core-processor-copper-ridge-multicore-hyperthreading,10482.html

Will be interesting to see how they compete.

i just want more competition in the market..it really sucks just having to see Sandy Bridge all the time..I want more choices!

Can't hate Intel for that. AMDs K10 help them back a bit as did their purchase of ATI.

 

[Derbixrace]

i think a cinebench R11.5 multithreaded run would have been a good test to show the capability of all cores of the CPUs

 

[jimmysmitty]

talking about intel and cloud. Mcafee Antivirus with a internet connection scored deep in the 90s. When the internet disappeared so did its anti malware abilities. Can see where's intel aiming for

I still am pretty much anti- McAffe, Norton, Trend Micro etc. Even if Intel owns them they are useless bloated software that doesn't do its job.

Too many PCs have I seen with that stuff on it with major viruses that McAffe will only see when MSE/MBam/SAS find it.

Back on topic, there are a lot of benchmarks I want to see BD on. But mainly I want a fair comparison. No Opteron dual socket vs desktop Core i7 single socket.