AMD - Piledriver chitchat

[watercooled]

Cheers, I've been looking for a die size comparison like that for ages.

[CAT-THE-FIFTH]

So the codename for the 2014 Jaguar successor has now been leaked:

http://www.fudzilla.com/home/item/30...essor-is-beema

[DanceswithUnix]

The change in the Phenom X4 line when it went from C2 to C3 stepping was quite noticable so that could be the only hardware change in Richland.

So pretty much what I said plus some firmware magic sauce - fair enough

If it is power management, it makes sense that the biggest improvements will be seen where power is at a premium - i.e. ULV mobile. I doubt this will improve on the 5800k that much, but imagine what the 5700 replacement will be like - the best desktop Trinity was undoubtedly the 5700, and that's the desktop product that will benefit most from the move to Richland? Can I get a Hell Yeah?!

This could make whatever replaces the 5500 pretty desirable too - and maybe we'll see 45W dual cores? Ooooh, so many possibilities...

But the figures AMD are quoting are CPU *or* GPU rather than game benchmarks which makes me think they are better able to balance energy from boosting CPU or GPU when the other is under light load. That could make the extra watts available in the 100W bin more valuable and make the top part stand out.

Better balance between CPU and GPU could make external graphics cards work better too.

[shaithis]

One of my mates has an FX6300 overclocked to 4.3GHZ and a HD5850 1GB. I have a Xeon E3 1220 quad core running at 3.1GHZ to 3.4GHZ with an HD5850 1GB too and in PS2,a game meant to run badly on AMD CPUs and is very CPU heavy, we were having a similar gaming experience. That was in online maps with hundreds of players,vehicles,explosions,etc and we running the game at 1920X1080 at high settings without motion blur and the complex shadows off,and the game seemed smooth enough. IIRC,the minimum framerates stayed consistently above 30FPS.

The funny thing is that after reading all the forums on how people needed overclocked Core i5 at 4GHZ and high end cards to run this game,even mates with lesser hardware were able to run it at lower settings fine.

From what I have seen so far, PS2 is a bit of a mess and needs seriously fixing/optimising. It's the only game I can remember where I get a smoother framerate out in the open with a gigantic draw distances, then a do in a building where I can't see further than 20 feet. Very odd indeed.

[CAT-THE-FIFTH]

A slightly excited article about HSA:

http://www.overclock.net/a/amd-fusio...-revolutionary

TH article about the history of AMD fusion and how it began:

http://www.tomshardware.com/reviews/...tory,3262.html

[Biscuit]

A slightly excited article about HSA:

http://www.overclock.net/a/amd-fusio...-revolutionary

TH article about the history of AMD fusion and how it began:

http://www.tomshardware.com/reviews/...tory,3262.html

Wow that article really is excited. All sounds promising for them if they can achieve it though.

[CAT-THE-FIFTH]

A range of CPUs tested with a GTX670:

http://translate.google.com/translat...omu-po-core-i7

The FX6200 does very well against the Intel Core i3 CPUs and considering that the FX6300 is generally faster for games,it is reinforces why I think it is a better CPU overall for a gaming rig.

[Eraovius]

A range of CPUs tested with a GTX670:

http://translate.google.com/translat...omu-po-core-i7

The FX6200 does very well against the Intel Core i3 CPUs and considering that the FX6300 is generally faster for games,it is reinforces why I think it is a better CPU overall for a gaming rig.

I agree 100% . Thats one of better CPUs price/performance. Core i3 is better only in few games, as Skyrim maybe, and Civilization V, that aren't so good-optimized imho. But 6300 Productivity and multithreading performance are almost 2 times better of
i3 3220

[DanceswithUnix]

Wow that article really is excited. All sounds promising for them if they can achieve it though.

Excited? It's like a puppy has scoffed a load of happy pills

I really liked:

"HSA is very innovative because aside from the benefits it will provide, it has been designed to succeed from the start."

like anyone makes an effort to design new tech to fail from the start...

[Biscuit]

haha i had the exact same thought!

AMD have made a bit of a reputation for building their next big thing up as a world changer and then it amounting to very little, so i will remain sceptical.

[DanceswithUnix]

haha i had the exact same thought!

AMD have made a bit of a reputation for building their next big thing up as a world changer and then it amounting to very little, so i will remain sceptical.

To be fair, I think they got quite a few right, they just aren't taken seriously until Intel start doing it as well.

Amusingly the only one that really stuck quickly was the AMD64 instruction set, and that is supposed to be only because Microsoft forced Intel to adopt it. That was kind of a game *not* changer Mind you, looking at the dead end alternative that is the Itanium architecture I am at least glad that didn't take off.

[scaryjim]

... looking at the dead end alternative that is the Itanium architecture I am at least glad that didn't take off.

Of course, if Itanium had taken off it probably wouldn't be a dead end now...

As to HSA, I think it's worth remembering - if I understand it correctly, anyway - that it won't magically make all current software 5x faster. Software will have to be developed specifically to benefit from it, and that's unlikely to happen (quickly) in the consumer market as AMD's market share is pretty small. After all, in the consumer market the software developers have no control at all over what hardware someone will try to run their software on.

It'll make a huge difference in the server and HPC markets though, where software development/procurement and hardware procurement are often intrinsically tied.

[DanceswithUnix]

Of course, if Itanium had taken off it probably wouldn't be a dead end now...

Itanium was always a stupid idea.
Intel is struggling to get more parallelism out of the awful architecture that is x86.
The answer is to schedule in blocks of 6 instructions? Now to run two bundles at once you have to compare all resources used in one bundle against all the resources used in the following bundles to run a bundle out of order. That's *really* hard. So you rely on software scheduling at compile time to get it right? Some things you only know at run time.

I just don't think Intel are very good at this stuff. The fact that people argue whether Intel or AMD are better when they both sell the same i86 junk, its like arguing whose farts smell best. Go sniff roses for pity's sake, there is better out there

[watercooled]

I wish someone would get AMD's VCE working properly in some transcoding software. AFAIK, even software which claims to use it now, in fact only uses OpenCL to make use of some GPU cores, not the dedicated hardware. Intel with their juggernaut marketing approach managed to get software out for it fairly quickly, but it seems people don't tend to code for specific features for AMD nearly as quickly.

I'm hoping, if the rumours are right about the next Xbox/PS using HSA, it will be a good, large-market learning tool for developers, so it can carry over to PC.

Even when Intel had a clear performance deficit to AMD for ~7 years, they still held something like 80% of the x86 market share. Combined with incredibly biased benchmarks commonly used by big companies today, you need more than a performance advantage to gain market share unfortunately.

[CAT-THE-FIFTH]

The thing is though that Adobe is starting to migrate over to OpenCL acceleration for their Creative Suite. The Mac version has already ditched CUDA and the PC version is going to probably do the same over the next year or so too.

The HSA Foundation does have some big backers though.



I will make a prediction. I expect Haswell to get better compute abilities and I expect "all of a sudden" more reviewers and forums will start to see how "great" using IGPs for acceleration is,if Intel does it.

[watercooled]

Agreed.

Somewhat off topic, but it seems Nvidia didn't bother with OpenCL/compute on Tegra4, and has lost customers because of it. Strange move for a company trying to push GPGPU?