Amd Officially Benches Bulldozer

[muziqaz]

Sparky, when comparing cpu effect you can use intels 2500K or even 2100 with 6870 vs some 4-6 core FX, and you will have same results, except this time the wouldn't be that stupid difference in prices.
I know everyone does these kinda tricks, but not at this magnitude. You can be silly a little bit and cheat in some components but not like this

[Spectrobozo]

Sparky, when comparing cpu effect you can use intels 2500K or even 2100 with 6870 vs some 4-6 core FX, and you will have same results, except this time the wouldn't be that stupid difference in prices.
I know everyone does these kinda tricks, but not at this magnitude. You can be silly a little bit and cheat in some components but not like this

indeed,
if you wanted, you could probably find games where a Celeron g530 can achieve the same performance as the "FX" in these GPU limited scenarios quite easily....

in my eyes this is only making AMD and the FX look bad, they should be comparing it to the similarly priced SBs on gaming, if that's what they want (compare CPUs when CPUs are not making the difference), or they should also use the 980x on their multi threaded comparison...

anyway, considering that the FX should scale much better than a CPU with HT, these multithreaded benchmarks on the slides are probably telling us that the FX will struggle to compete with SB in CPU limited gaming and many other 1-4 threads tasks.... so I wouldn't be surprised to see a cheaper i3 2100 beating the FX in many games, so AMD claiming a hugely lower cost for the same gaming performance looks really.....

[gosh]

Sparky, when comparing cpu effect you can use intels 2500K or even 2100 with 6870 vs some 4-6 core FX, and you will have same results, except this time the wouldn't be that stupid difference in prices.
I know everyone does these kinda tricks, but not at this magnitude. You can be silly a little bit and cheat in some components but not like this

These slides are probably not made for xtreme members. How many out there knows what a CPU is?

[Movieman]

and perhaps the purpose of these slides was to show total system performance at a given price point yes?
Saying maybe that" We can match the other guys without having to spend the BIG money on that monster $1000.00 cpu
or if your budget has the money to include that monster cpu why not go this way and add a second vid card with that extra money.
That's how I read these slides.
Of course, what the hell do I know about gaming

[muziqaz]

Movieman, nothing?
But then again, if you have so much money for $1k CPU you can spend less on even cheaper intel cpu and have the same results.
Anyway, I expressed my opinions on this matter, I will shut up now

[Movieman]

Movieman, nothing?
But then again, if you have so much money for $1k CPU you can spend less on even cheaper intel cpu and have the same results.
Anyway, I expressed my opinions on this matter, I will shut up now

Yes and no issues..Opinions will vary and all are welcome.
What I see here is that now on the lets call it the normal dollar range there are choices that weren't available a few months back.
Like I said in another thread, this will be a good year for all of us as both companies are bringing out excellent products.

[muziqaz]

<..>
What I see here is that now on the lets call it the normal dollar range there are choices that weren't available a few months back.
Like I said in another thread, this will be a good year for all of us as both companies are bringing out excellent products.

Agreed

[Dimitriman]

It is indeed stupid to use a 6870 to present CPU limitations on gaming benchmarks. But then again I expect nothing less from PR slides.
As to being fake, I am not sure, but if Bulldozer lands between 2500k and 2600k overall I would not call that fail. Fail would be slower or equal to 1100T.

There will be room to develop this new core. I see most of its weakness being IPC performance, with the proper tweaks it can improve greatly.

[P.A.D>]

I spent this much money (1000$) on a CPU once and felt like a complete fool after doing so, since I like many others OC for fun, and most of all to save money for other parts as Movieman says.

[Hornet331]

It is indeed stupid to use a 6870 to present CPU limitations on gaming benchmarks. But then again I expect nothing less from PR slides.
As to being fake, I am not sure, but if Bulldozer lands between 2500k and 2600k overall I would not call that fail. Fail would be slower or equal to 1100T.

There will be room to develop this new core. I see most of its weakness being IPC performance, with the proper tweaks it can improve greatly.

Well we haven't seen many facts till now...

How about clock to clock comparison between all available cpus
How high can i go on the FX with air/water cooling (extreme cooling is another topic )

I am waiting for full reviews before making any further statements.

[Sparky]

Sparky, when comparing cpu effect you can use intels 2500K or even 2100 with 6870 vs some 4-6 core FX, and you will have same results, except this time the wouldn't be that stupid difference in prices.
I know everyone does these kinda tricks, but not at this magnitude. You can be silly a little bit and cheat in some components but not like this

See, I KNEW I was missing something. Dumb not-yet-awake posts

[rintamarotta]

Remember all, these slides are ment for people who are clueless about GPU or CPU limitations.

[flyck]

The WPrime score seems low for the 2600K, the % difference between the 2500K and 2600K should be 40+%:
http://www.bit-tech.net/hardware/cpu...ridge-review/7
http://www.hardocp.com/article/2011/...ssors_review/3

The difference in the HardOcp link is 29%, so far away from 40%+

[accord99]

The difference in the HardOcp link is 29%, so far away from 40%+

The proper way to calculate how much faster the 2600k is versus the 2500k and compare with the values given in the slide, is for the 2500k time to be divided by 2600k time. This leads to the 2600k being 42.6% faster.

[flyck]

Then you say the 2500 is 42.6% slower. But when the baseline is the 2500 you say the 2600 is 29% faster. In those slides the baseline is 2500 so in that relation the 2600 is 29% faster. If the baseline was the 2600 the 2500 would be 42.6% below that.

[accord99]

Then you say the 2500 is 42.6% slower. But when the baseline is the 2500 you say the 2600 is 29% faster. In those slides the baseline is 2500 so in that relation the 2600 is 29% faster. If the baseline was the 2600 the 2500 would be 42.6% below that.

Because the WPrime score is given in seconds and where lower is better, you use the inverse of what you are suggesting. Or you can think of it this way, the 2500K does 32M digits in 10.37 seconds, or the equivalent of 3.086 million digits per second. The 2600K does 4.4 million digits per second or 42.6% more work (faster) than the 2500K.

[ThePointer]

Then you say the 2500 is 42.6% slower. But when the baseline is the 2500 you say the 2600 is 29% faster. In those slides the baseline is 2500 so in that relation the 2600 is 29% faster. If the baseline was the 2600 the 2500 would be 42.6% below that.

I have no intention to enter discussion about the topic but only touch the graph representation

if the graph wish to show longer bar represent better performance, and the data from other site such as http://www.hardocp.com/article/2011/...ssors_review/3 using shorter representing faster, they could be converted by simply invert them

2500K - 10.37s - 0.09643 (1/10.37)
2600K - 7.27s - 0.1376

if 2500K is normalized to 1
then
2500k = 0.09643/0.09643 = 1
2600 = 0.1376/0.09643 = 1.427

so in terms the bar representation, 2600K should be at 143%

[Lanek]

A little question about FMA4 and XOP instructions set ( AVX-AMD Intel use FMA3 and not the same AVX ( no XOP )... Can thoses different instruction set have an impact somewhere if the software, or codes, use a specific compiler ( from intel or AMD ).

I think specially on some synthetic benchmarks as SandraSys, where the benchmarks use FMA and AVX instructions, but i think we can count easely see them in "real " appliccations. Otherwise, does the softwares code need be updated or optimised taking in count thoses different instructions set ?

[rcofell]

Because the WPrime score is given in seconds and where lower is better, you use the inverse of what you are suggesting. Or you can think of it this way, the 2500K does 32M digits in 10.37 seconds, or the equivalent of 3.086 million digits per second. The 2600K does 4.4 million digits per second or 42.6% more work (faster) than the 2500K.

Just to formalize what you two are at:

• Speedup = Time_Old / Time_New
• Time = CPI * Instr_in_prog / Clk_freq
  (CPI = cycles per instruction, the inverse of IPC)

A little question about FMA4 and XOP instructions set ( AVX-AMD Intel use FMA3 and not the same AVX ( no XOP )... Can thoses different instruction set have an impact somewhere if the software, or codes, use a specific compiler ( from intel or AMD ).

I think specially on some synthetic benchmarks as SandraSys, where the benchmarks use FMA and AVX instructions, but i think we can count easely see them in "real " appliccations. Otherwise, does the softwares code need be updated or optimised taking in count thoses different instructions set ?

To start from the beginning, additions to the ISA are will most-always require updates to the compiler and hence re-compile of the code. By using an applicable optimizing compiler (like Intel's ICC or AMD's x86 Open64), just compiling with the new instruction MIGHT be enough to exploit the instruction set well enough, heavily depending on how the original code was written.

I'm not quite an expert when it comes to modern compilers and what commonly is necessary for the programmer to prod at... but in the case of FMA4 (and FMA3), the target programs should at least already be written using intrinsics (a way to express SIMD structures to the compiler) and already have the related operations part of the algorithm. This is more applicable to software using image/signal processing methods.

To summarize, compilers that are updated to use (and hopefully optimize) for the new instructions are all that's needed to use them with your code. However, it's usually the case that the code has to be written (from an algorithmic and somewhat syntax perspective) such that the compiler knows what to do.

[flyck]

Because the WPrime score is given in seconds and where lower is better, you use the inverse of what you are suggesting. Or you can think of it this way, the 2500K does 32M digits in 10.37 seconds, or the equivalent of 3.086 million digits per second. The 2600K does 4.4 million digits per second or 42.6% more work (faster) than the 2500K.

correct.

to myself...
to my other self..

Seems monday is a day to stay away from the send button.

[god_43]

...where is interlagos, i thought it was being released today?

[muziqaz]

...where is interlagos, i thought it was being released today?

1st thing, are you trolling?
2nd thing: where did you get this date?
3rd thing: interlagos started shipping not too long ago with release date in q4(it was announced officially by John Fruehe).
4th thing: where do you want to find those interlagos? it's a server CPU. Do you expect to go in into BestBuy and buy one off the shelf?
And finally, who cares when interlagos is released. Main thing is it has shipped and is being validated and fitted into server systems. and those companies will release their systems after platform gets validated.
For all we know AMD could come out and say, we just released interlagos, but it is now not up to us, but for system builders. Would it make any difference to you?
Unless you want to do some freakish calculations and extrapolate FX line up performance out of some 16 core 1.9ghz Interlagos chip with server OS and low clocked DDR3.

[EniGmA1987]

1st thing, are you trolling?
2nd thing: where did you get this date?
3rd thing: interlagos started shipping not too long ago with release date in q4(it was announced officially by John Fruehe).
4th thing: where do you want to find those interlagos? it's a server CPU. Do you expect to go in into BestBuy and buy one off the shelf?
And finally, who cares when interlagos is released. Main thing is it has shipped and is being validated and fitted into server systems. and those companies will release their systems after platform gets validated.
For all we know AMD could come out and say, we just released interlagos, but it is now not up to us, but for system builders. Would it make any difference to you?
Unless you want to do some freakish calculations and extrapolate FX line up performance out of some 16 core 1.9ghz Interlagos chip with server OS and low clocked DDR3.

WTF kind of response is this? No he isnt trolling, just asking a question. It was around the internet that Sept. 26th was the official launch for Interlagos. Just cause system builds can use it doesnt mean you also cant find the processor at a etailer to buy. I am waiting for one as well for a server I am building. Also, you can use 1600 speed DDR3 which isnt exactly "low clocked" IMO. And you clearly dont even know the models since there is no 1.9GHz model, the slowest speed Interlagos starts at 2.1GHz base with a 2.6GHz turbo. Dont post stupid stuff if you dont know what your talking about.

[EniGmA1987]

1st thing, are you trolling?
2nd thing: where did you get this date?
3rd thing: interlagos started shipping not too long ago with release date in q4(it was announced officially by John Fruehe).
4th thing: where do you want to find those interlagos? it's a server CPU. Do you expect to go in into BestBuy and buy one off the shelf?
And finally, who cares when interlagos is released. Main thing is it has shipped and is being validated and fitted into server systems. and those companies will release their systems after platform gets validated.
For all we know AMD could come out and say, we just released interlagos, but it is now not up to us, but for system builders. Would it make any difference to you?
Unless you want to do some freakish calculations and extrapolate FX line up performance out of some 16 core 1.9ghz Interlagos chip with server OS and low clocked DDR3.

WTF kind of response is this? No he isnt trolling, just asking a question. It was around the internet that Sept. 26th was the official launch for Interlagos. Just cause system builds can use it doesnt mean you also cant find the processor at a etailer to buy. I am waiting for one as well for a server I am building. Also, you can use 1600 speed DDR3 which isnt exactly "low clocked" IMO. And you clearly dont even know the models since there is no 1.9GHz model, the slowest speed Interlagos starts at 2.1GHz base with a 2.6GHz turbo. Dont post stupid stuff if you dont know what your talking about.

[Manicdan]

hahahaha, theres lots of people around here who use server chips, and some of those people are Mods.