Llano Demo

[drizek]

I'm somewhat disappointed in myself for not making the connection on another amusing point of this demonstration earlier.

How exactly did AMD put a FCPGA998 i7-2630QM chip into the LGA1155 socket of the Asus P8H67-M that they claim to have used?

Technically, they never claimed to have used the Asus motherboard, just the chipset from that motherboard.

It's possible that they underclocked a desktop chip though.

 

[busydude]

Technically, they never claimed to hace used the Asus motherboard, just the chipset from that motherboards.

They mentioned it in the specs.. if they had under-clocked a desktop chip, they should have atleast mentioned it. Khato got a good point though.. the only thing that I can think of is, they made an error.. it is certainly possible.

 

[Concillian]

Every thread were grammer cops enter I just by instict get worserer and worserer . Its funny not that long ago a gent here did his best at bad spelling and grammer yet everone could understand every word he wrote perfectly. Grammer spelling cops are just diggs.

Nothing in my posts was a comment on grammar, or could even be construed that way. Not sure how you read that post and figured I was making a comment on grammar. You brought that up yourself, is there a reason you're particularly defensive about that?

No, my comments are on how you have started with pretty valid points, but have slowly taken over this thread with nonsense that does not relate to the original points you made, and are quite far from the original topic of the thread. In my last post, I demonstrated exactly how you did it in one set of posts in the discussion. It's not the only place it has happened in this thread, and the end result is a total mess that has little value to anyone.

You have valid points about how the comparison was done, however your methods remove value from a lot of what you say. You come off as a foaming at the mouth fanboi, and people naturally will dismiss the opinions of people like that. My call for moderation is not because your opinions are invalid, but because of the way you present them in every Zacate and Llano thread on this forum. That's exactly what moderation is for, not to influence opinion, but to guide it so that it is expressed in a way that's constructive instead of destructive.

 

[Nemesis 1]

They mentioned it in the specs.. if they had under-clocked a desktop chip, they should have atleast mentioned it. Khato got a good point though.. the only thing that I can think of is, they made an error.. it is certainly possible.

Well AMD seems to make alot of errors when testing against Intel this is 2 in a row .

 

[Nemesis 1]

Nothing in my posts was a comment on grammar, or could even be construed that way. Not sure how you read that post and figured I was making a comment on grammar. You brought that up yourself, is there a reason you're particularly defensive about that?

No, my comments are on how you have started with pretty valid points, but have slowly taken over this thread with nonsense that does not relate to the original points you made, and are quite far from the original topic of the thread. In my last post, I demonstrated exactly how you did it in one set of posts in the discussion. It's not the only place it has happened in this thread, and the end result is a total mess that has little value to anyone.

You have valid points about how the comparison was done, however your methods remove value from a lot of what you say. You come off as a foaming at the mouth fanboi, and people naturally will dismiss the opinions of people like that. My call for moderation is not because your opinions are invalid, but because of the way you present them in every Zacate and Llano thread on this forum. That's exactly what moderation is for, not to influence opinion, but to guide it so that it is expressed in a way that's constructive instead of destructive.

The grammer cop thing wasn't for you . I just put it into that post as others mentioned it .

Iam not crapping on llano or zacata topics . It is AMD crapping on intel . If llano is all that. AMD can do a C2D review like intel did . Zacata is going into tablets according to some here . AMD can just show their product performance rather than creating hype about a product that they compare againt OLd atoms and outdated desktop parts. When you choose to compare a product against the comp. At least have the balls to do more bench marks than simplely crippling a system to make self lok good. I remember bach when AMD had the lead and publicly challenged intel to a show down . Intel didn't get into a pissing match with them . Intel simply showed c2d 6 monts befor release and killed AMD sales and stock price. AMD also pissed on everone in the industry when Sanders said real men have Fabs . Weres those AMD fabs now . Were is the pissing contest AMD was itching for. This is not a llano only topic as SB was an understudy in this star packed thriller

 

[bunnyfubbles]

I'm sceptical about this demo. SB obliterates the "stars" CPU core, and its integrated graphics are pretty good as well. AMD has definitely rigged this test.

AMD needs to come out with their Bulldozer core with a good integrated GPU.

wtf for, this is clearly a demonstration to show Llano's competitiveness in a laptop / low power environment, Bulldozer is a completely different animal targeting higher end desktop rigs where discrete GPUs are inferred.

 

[busydude]

wtf for, this is clearly a demonstration to show Llano's competitiveness in a laptop / low power environment, Bulldozer is a completely different animal targeting higher end desktop rigs where discrete GPUs are inferred.

Next gen Llano is going to have BD modules instead of Stars core. This gen Llano is going to be last CPU based on Stars core.

 

[Concillian]

Next gen Llano is going to have BD modules instead of Stars core. This gen Llano is going to be last CPU based on Stars core.

"Next get Llano" will not be Llano.

Llano uses stars core.
"Next gen Llano" will use BD modules, but is going to be at least a year and will be called something other than Llano. It's fun to speculate, but really a lot will change between now and then.

There is a lot of time for things to happen like:
- Intel has time to fix it's drivers
- Intel can fix the atrocious chipset situation so that EUs can be used at the same time as discrete graphics in all chipsets.
- Software developers will have time to develop applications that use the GPU portion of CPUs for more than just low quality gaming.

Right now GPU on CPU matters for primarily a narrow (but large volume) market segment: those who want to do casual gaming without a discrete GPU, either due to form factor (all-in-one / notebook), price, or aesthetics (small PCs). Since gaming is the primary application for on-die GPU right now, drivers are so important that AMD should be able to pick up marketshare just from compatibility, since Intel video drivers have such a poor reputation. Also, in this low of a gaming performance category, CPU speed is not so important, and the performance of the CPU side doesn't hurt so much.

By the time BD module based APUs are out, there is plenty of time for the landscape to change dramatically. Ivy Bridge should be out too, Intel will have a generation to correct many of it's own mistakes. Intel probably has a fair bit of performance to gain from driver maturity as well. Plus more applications than games will start using the on-die GPU for things other than games.

 

[Arkadrel]

Since gaming is the primary application for on-die GPU right now,

Give it a years time, and we ll see APUs use their GPUs for photoshop, video editing, Internet Explore web browseing, movie decodeing ect.

Intel will need to figour out how to have Ivy bridge do GPGPU via OpenCL or DirectCompute or its gonna fall behinde in compute power.


Does anyone know if Sandy Bridge can do OpenCL/DirectCompute/APP/CUDA? Or if Ivy Bridge will be able too?

 

[busydude]

"Next get Llano" will not be Llano.

Yes.. and the code name is Trinity. I made that post in a hurry.. so could not remember the exact code name.

Llano uses stars core. "Next gen Llano" will use BD modules, but is going to be at least a year and will be called something other than Llano. It's fun to speculate, but really a lot will change between now and then.

 

[Nemesis 1]

Give it a years time, and we ll see APUs use their GPUs for photoshop, video editing, Internet Explore web browseing, movie decodeing ect.

Intel will need to figour out how to have Ivy bridge do GPGPU via OpenCL or DirectCompute or its gonna fall behinde in compute power.


Does anyone know if Sandy Bridge can do OpenCL/DirectCompute/APP/CUDA? Or if Ivy Bridge will be able too?

Well I won't say on this one but if we pay close attention to Apple who is the father of Open CL which I am sure intel had nothing to do with. So the first sigms will come from apple I like Apple and intel working to gether. I never understood the whole wintel thing . As IBM was a big player in Intels success. But the Dog that was trained to take out DEC turned on his master.

 

[Concillian]

Give it a years time, and we ll see APUs use their GPUs for photoshop, video editing, Internet Explore web browseing, movie decodeing ect.

Isn't that what I said in the rest of my post?

By the time BD module based APUs are out, there is plenty of time for the landscape to change dramatically. Ivy Bridge should be out too, Intel will have a generation to correct many of it's own mistakes. Intel probably has a fair bit of performance to gain from driver maturity as well. Plus more applications than games will start using the on-die GPU for things other than games.

I thought it was so important that I apparently said it twice in one sentence ... lol.

 

[Cerb]

The 64 x86 thing . To compare itanic to a 64bit x86 after the fact is way off the mark . Had hammer not happened Intel wouldn't be making X86 now . EPIC is were intel was heading . Doesn't BD now contain elements that went into itanic .

Tens of MB caches, massive binaries, piss poor peformance in code with tons of indirect branches straight out of a standard compiler, static scheduling, putting tons of their best and brightest on the porject in a final effort to make it profitable (successful), once it became obvious that compilers weren't there yet?

I don't see any of that in BD, or for that matter, any other modern general-purpose CPU. No doubt Itanium offers excellent R&D results, and has probably paid for itself in x86 developments, but from the ISA on up, it's not a good general-purpose CPU. The idea that hardware is inferior to software in terms of scheduling and prediction requires either (a) additional instructions doing speculation and on-CPU memory management running as part of every thread, and/or (b) assuming that unique inputs do not create unique outputs, regarding good branch prediction and memory fetching formulas and algorithms (which would allow them to be added to IA64 CPUs, to provide a, "have your cake, and eat it too," result, if it weren't generally false).

It's not just 64-bit addressing, but that the Itanium would have needed to have much lower power consumption, and much lower cost, with no detriment to integer performance, for it to not fall flat as a mainstream CPU. The Itanium requires specific optimizations per application to perform decently, also, if you don't get lucky compiling it, which is not something mainstream developers will do. Some will be willing to have someone go in and take it from performing OK to performing great; but most need to be able give the code to something like GCC, MSVCC, etc., and use the result. The K8 took IA64's mainstream adoption from being a wasting disease to being smacked by a truck. Even w/o the K8, it still would not have happened.

EPIC might have been where Intel was heading, but the rest of the world wasn't. EPIC for certain tasks, like heavy duty number crunching, works pretty well. Most people don't do much of that (at least, not at the cost of other workloads), and hardware scheduling and speculation still haven't reached their limits, by a long shot. IA64 for the mainstream was no more a good idea than a hundreds-of-watts 10GHz desktop CPU was. Intel was in its own little world, acting as though the wants and needs of computer buyers were irrelevant.

 

[Concillian]

Does anyone know if Sandy Bridge can do OpenCL/DirectCompute/APP/CUDA? Or if Ivy Bridge will be able too?

The hardware can do at least some of those, it's just a matter of the drivers / software layer, which I don't think supports any of those at this point in time. Pretty sure they can add these features later with driver updates.

The real problem with SB is that a discrete card essentially completely shuts off the GPU portion of SB on P67, so a significant portion of the enthusiast market is completely shut-out from any benefits that come in this area.

Due to this short-sightedness on the launched SB chipsets, I don't expect to see Intel really push the envelope on these kinds of applications of on-die GPUs. I expect them to sit back, let things develop and will support whatever happens to end up to be popular. By then Ivy Bridge will be out and their lack of foresight with P67 will be far enough behind that it won't draw attention.

 

[Arkadrel]

The real problem with SB is that a discrete card essentially completely shuts off the GPU portion of SB on P67, so a significant portion of the enthusiast market is completely shut-out from any benefits that come in this area.

1) Intel need to figour out how to use gpu resources of the sandy bridge gpu, even when a discrete put in.

2) Intel will need a version of hybrid crossfire, that ll work with nvidia/amd/intel? discrete cards.

3) Intel will need to get GPGPU capabilities on IGP's (that work with discrete cards too)

4) Intel will need to get full directx 11 support without any work arounds.

5) Intel Ivy bridge will need a 300% increase in GPU speeds to compete with the Llano's GPU.

6) Intel needs driver quality to match AMDs + something simular to MLAA that could be big for the low end IGPs.


^just assumeing above, because "intel + disadvantages" dont go together, in the same sentense.

Its alot of hurdels.....Im sure Intel has enough engineers to solve the issues above though.

 

[grimpr]

So the first sigms will come from apple I like Apple and intel working to gether.

Bookmarked, get ready for big sensations and vibrations...

 

[itsmydamnation]

Oh my I be so sorry I didn't put the QM on the 2820 I caught it right off but assumed a smart guy like you would have no problem .

no shit sherlock, its not in the list,
http://www.anandtech.com/bench/CPU/2
but again whats your actual point, what are you comparing to what, or which set of goal posts have you moved this time. Are you actually going to answer any questions ( i guess not, trolls dont tend to do that).


edit:
no form of AA will be big on APU's, AA is a bandwidth kill no matter which way you look at it.

 

[Arkadrel]

no form of AA will be big on APU's, AA is a bandwidth kill no matter which way you look at it.

It uses very little resources, which is perfect for IGPs.



No MLAA (say Intel IGP): From FFXI:

AMD IGP (with MLAA on): From FFXI:

I think MLAA could be a selling point for AMD IGPs.
It certainly would make a Visual differnce if the alternative is no AA because you dont have the resources to use it.

 

[Idontcare]

no shit sherlock

Please be mindful of the fact that we don't allow profanity in the technical forums. Try and keep it clean. (posting as member, not as mod)

 

[itsmydamnation]

Please be mindful of the fact that we don't allow profanity in the technical forums. Try and keep it clean. (posting as member, not as mod)

sorry, a very frustrating customer to deal with.....

 

[IntelUser2000]

re run the test with IGP locked at default 850mgz than tell us the score

Just for curiosity, I ran the same thing with the GPU now at 850MHz. So Turbo shouldn't kick in at all(if for some obscure reason having Turbo off still activated GPU Turbo).

Score of 51, which contrasts to 56 I got at 1GHz. It must be bound by something else beyond 1GHz.

 

[IntelUser2000]

I'm somewhat disappointed in myself for not making the connection on another amusing point of this demonstration earlier.

How exactly did AMD put a FCPGA998 i7-2630QM chip into the LGA1155 socket of the Asus P8H67-M that they claim to have used?

From the screen size it sounds like they are doing the test on some sort of a test bed, rather than on an actual system. Initially I thought it was a notebook with 14 inch screen, but how common are they nowadays, and finding an identically matched laptop model is hard to do. Plus, Llano might be only in testing stage at the moment.

 

[Idontcare]

From the screen size it sounds like they are doing the test on some sort of a test bed, rather than on an actual system. Initially I thought it was a notebook with 14 inch screen, but how common are they nowadays, and finding an identically matched laptop model is hard to do. Plus, Llano might be only in testing stage at the moment.

With the prevalence of Intel systems on the market, they are going to succeed in doing little but gaining scrutiny and suspicion by using/creating a non-standard testbed for their demo.

They really ought to know better, perception of fairness and no shenanigans is paramount in side-by-side comparisons. There is really no excuse for them to have not used off-the-shelf retail available Intel kit.

 

[chewietobbacca]

Well I won't say on this one but if we pay close attention to Apple who is the father of Open CL which I am sure intel had nothing to do with. So the first sigms will come from apple I like Apple and intel working to gether. I never understood the whole wintel thing . As IBM was a big player in Intels success. But the Dog that was trained to take out DEC turned on his master.

Interestingly enough, Apple moved their MacbookPro's to ATI GPU's

 

[Nemesis 1]

Just for curiosity, I ran the same thing with the GPU now at 850MHz. So Turbo shouldn't kick in at all(if for some obscure reason having Turbo off still activated GPU Turbo).

Score of 51, which contrasts to 56 I got at 1GHz. It must be bound by something else beyond 1GHz.

Thanks . I wanted to see how close it would get you to the AMD video score . Now if you could get down to 2ghz on the cpu I think we might be close to that score .

Thanks for the time and effort . Your a basic good person.