Found another one: Ukraine store accidentally ships FX-8120 and it gets tested!

[-Slacker-]

^My fingers are crossed Well, not really, that would be silly

Anyway, I'm still hoping someone would tackle this:

What I'm trying to get at is that, maybe, the 300w tdp could be due to the stock cooler not keeping the cpu cool enough at that voltage/wattage (remember that tdp goes up if the die temperature rises).

post #191

 

[Tsavo]

Single channel ram?

 

[aphelion02]

You think that this is a point ??..

I can tell you that next year s 220$ chips will deliver , historically ,
the best perfs/price performance ever for that category of prices...

Are you truly this thick? Its obvious he meant that not absolute performance but performance relative to the competition at the time. the 2500k is the 8800GT of GPUs.

 

[Skurge]

Ya, with 3 days until launch, you might as well wait.

Curious as to why you like the 990FX platform more. Its SATA 3 performance with an SSD is a lot slower.

I've got these 2 boards.

http://www.newegg.com/Product/Produc...82E16813128508

http://www.newegg.com/Product/Produc...82E16813128499

The 990FX is $100 less.

The slower Sata3 performance is another thing to consider. I'll have to look at some more reviews and think about it.

 

[bunnyfubbles]

I've got these 2 boards.

http://www.newegg.com/Product/Produc...82E16813128508

http://www.newegg.com/Product/Produc...82E16813128499

The 990FX is $100 less.

The slower Sata3 performance is another thing to consider. I'll have to look at some more reviews and think about it.

I'm in love with Z68 if only for SSD caching

128GB Crucial M4 SSD for my OS and most important apps/games

64GB Microcenter rebranded ADATA Sandforce SSD accelerating a 1TB Samsung F3 HDD for my huge collection of Steam games and other games that I don't want/need taking up space on my M4

3 x 1.5TB Hitachi 7K3000 HDDs in RAID0 for video capture

makes it even harder for AMD to compete with as a gaming platform IMO

 

[Abwx]

Are you truly this thick? Its obvious he meant that not absolute performance but performance relative to the competition at the time. the 2500k is the 8800GT of GPUs.

Do you think so ??..

At the time , according to his source , a pentium D 2.8ghz was 280$ ,
to be compared to a 2 ghz A64 X2 that was 354$..

That pretty much hold the comparison with the SBs current pricing
compared to the competition , and is even better considering that
in MT a 2600K is nowhere more performant than a X6 in term of the
same price/perfs comparison as their old siblings..

 

[Silverforce11]

When is Pile Driver due, Q1 2012?

I hope they fix whatever is wrong with BD, cos it looks like a turd about to get stomped on release.

Edit: Would perf numbers differ with Win7 SP1? Looks like lab501.ro used Win7 7600 build.

 

[NostaSeronx]

When is Pile Driver due, Q1 2012?

I hope they fix whatever is wrong with BD, cos it looks like a turd about to get stomped on release.

Edit: Would perf numbers differ with Win7 SP1? Looks like lab501.ro used Win7 7600 build.

Trinity is do 1H 2012
Vishera is do Q2-2H 2012

Edit: Would perf numbers differ with Win7 SP1? Looks like lab501.ro used Win7 7600 build.

No

Software Recompile isn't needed

Pardon the big image

Interlagos vs i7 2600K wow AMD

 

[inf64]

@NostaSeronx
That is 8C SB-E in the slide above. You can see the core counts from the very slide: MC's Flops are 4x12 ,Westmere's 4x6,Interlagos' 4x16(fmacs) or 8x8(FLexFPs), SB-E's 4x8 or 8x8 . It's very simple.

 

[nyker96]

after looking at the preview benches, i have to wonder, what have they been doing for 5 years in R&D. i mean 8 core v. 4 core, it should be at least that much better in multithreaded, but even in threaded apps, it's just soso compare to 2600k. This cannot be. very disappointed at bd, i have thought it won't catch up to 2600k but at least very close. sigh. they need better r&d is what it is.

 

[Tsavo]

after looking at the preview benches, i have to wonder, what have they been doing for 5 years in R&D. i mean 8 core v. 4 core, it should be at least that much better in multithreaded, but even in threaded apps, it's just soso compare to 2600k. This cannot be. very disappointed at bd, i have thought it won't catch up to 2600k but at least very close. sigh. they need better r&d is what it is.

They've been getting rid of chunks of their R&D department for the better part of a decade.

 

[bronxzv]

@NostaSeronx
That is 8C SB-E in the slide above. SB-E's 4x8 or 8x8 . It's very simple.

simple but wrong, a SNB core can do 16 single precision FLOP per clock with AVX-256 (1 vmulps + 1 vaddps per clock), it should read 4x16, 6x16 or 8x16 for SNB-E

 

[inf64]

SB-E can't do 2x256bit loads per cycle so what you wrote is just theoretical throughput. The slide is misleading in other aspect though: Opteron 6100 can do 8 SP FLOP operations per cycle per core,not 4 (for the same reason since it has 1x128it Fmul and 1x128bit Fadd pipeline per core)

 

[bronxzv]

what you wrote is just theoretical throughput.

well, this slide is about theoretical throughput

 

[Black96ws6]

Watch the benchmarks live:

http://www.justin.tv/bassbench#/w/1883595104

 

[RussianSensation]

The AMD stock cooler is much better than Intel's junk.

The tin can packaging is also quite nice.

What if AMD FX performs 20-30% faster in games/Eyefinity when paired with an AMD GPUs?

 

[NostaSeronx]

Zambezi FX-8150
4x8 (128bit for 8 units) => 32 Flop
8x4 (256bit for 4 units) => 32 Flop
Sandy Bridge i7 2600K
4x8 (128bit for 8 units) => 32 Flop
8x8 (256bit for 8 units) => 64 Flop

FX-8150 has 4 256bit FPs
i7 2600 has 8 256bit FPs
It's not hard to notice that

Xeon Esomething(8 core SB-E)
4x16 (128bit for 16 units) => 64 Flop
8x16 (256bit for 16 units) => 128 Flop

 

[Abwx]

simple but wrong, a SNB core can do 16 single precision FLOP per clock with AVX-256 (1 vmulps + 1 vaddps per clock), it should read 4x16, 6x16 or 8x16 for SNB-E

Granted this is possible , how much would be the ratio
of AVX in a compiled code ?...
I doubt it would be more than a handfull of %...

Indeed , an interessant discussion about it , isnt it ?..

http://software.intel.com/en-us/forums/showthread.php?t=68554

 

[Black96ws6]

Watch the actual SCREENS here!:

https://join.me/743-658-418

 

[bronxzv]

Granted this is possible , how much would be the ratio
of AVX in a compiled code ?...
I doubt it would be more than a handfull of %...

a properly optimized application will use AVX-256 for all hotspots, it may be only 5% of the total code but 50% of the retired instructions since you spend most of the runtime in a few well optimized inner loops

Indeed , an interessant discussion about it , isnt it ?..

http://software.intel.com/en-us/forums/showthread.php?t=68554

ah, nice memories, my code was ported to AVX-256 more than one year before Sandy is available

 

[LondonBurning]

What if AMD FX performs 20-30% faster in games/Eyefinity when paired with an AMD GPUs?

This is what I'm curious about. We'll see...

 

[Black96ws6]

New link, use this one:

https://join.me/573-962-519

 

[jones377]

Zambezi FX-8150
4x8 (128bit for 8 units) => 32 Flops
8x4 (256bit for 4 units) => 32 Flops
Sandy Bridge i7 2600K
4x8 (128bit for 8 units) => 32 Flops
8x8 (256bit for 8 units) => 64 Flops

FX-8150 has 4 256bit FPs
i7 2600 has 8 256bit FPs
It's not hard to notice that

Xeon Esomething(8 core SB-E)
4x16 (128bit for 16 units) => 64 Flops
8x16 (256bit for 16 units) => 128 Flops

You're still wrong, a SB core can do 2*256bit AVX FP per cycle. I'm also curious to how you derive flops without clockspeed? (hint, you cant) Note that in AVX optimized Linpack, 4C SB has already gotten above 100GFLOPS

 

[NostaSeronx]

You're still wrong, a SB core can do 2*256bit AVX FP per cycle. I'm also curious to how you derive flops without clockspeed? (hint, you cant) Note that in AVX optimized Linpack, 4C SB has already gotten above 100GFLOPS

The units are their 8 units = 2*256bit and 2*128bit per core

4 = 4*32bit operands
8 = 8*32bit operands

Operands x Units = Flop

 

[Abwx]

I'm also curious to how you derive flops without clockspeed?

One flop , two flops....