The price differential between 4 GB of GDDR5 and 4 GB of DDR3 is probably in the range of $40-60, so it's not a whole lot different from say SSD over hard drives, or IPS panels.
They certainly aren't being clear.
But let's face it, what manufacturer is going to put GDDR5 system memory in these? They would price themselves completely out of the market.
they would price in the same market as low end discreet cards...
I don't want to type it all out again (search my post history for a full description of why if you want it), but this is totally incorrect. GDDR5 doesn't have higher latency than DDR3. In fact, GDDR5 has 0 (zero) performance disadvantages when compared to DDR3. GDDR5 is strictly superior to DDR3 when it comes to performance, and is only ill-suited to being main system memory for economic, serviceability, and manufacturing reasons.
Not sure but seems like it.But I really wanna know the programming model becasue some wild pointer quirks can halt this system.
So they just said that there will be GDDR5 APUs...
The impression I got was that he was talking about discrete GPUs specifically since their memory has to be accessed over the PCIe bus. A different quote mentions being able to choose either DDR3 or GDDR5. I wasn't sure if that meant you can mix the two. I could envision a scenario in which you'd buy, for example, a motherboard with 2GB of GDDR5 soldered to it and then add extra DDR3 like normal. The amount of GDDR5 onboard would make for a nice product differentiator, which the motherboard manufacturers would probably like.That makes it highly doubtful that GDDR5 will be used for system memory. They are even flat out saying it won't be, so at least we can put that rumor to rest.
So, as the article says - and I didn't catch this either - unified memory space doesn't mean unified memory chips.
I think that would only work if they are the same voltage. Are they? (I somehow think mem volts on high-performance graphics cards are a bit lower than 1.5v, aren't they?)I could envision a scenario in which you'd buy, for example, a motherboard with 2GB of GDDR5 soldered to it and then add extra DDR3 like normal.
I think that would only work if they are the same voltage. Are they? (I somehow think mem volts on high-performance graphics cards are a bit lower than 1.5v, aren't they?)
I could envision a scenario in which you'd buy, for example, a motherboard with 2GB of GDDR5 soldered to it and then add extra DDR3 like normal. The amount of GDDR5 onboard would make for a nice product differentiator, which the motherboard manufacturers would probably like.
The impression I got was that he was talking about discrete GPUs specifically since their memory has to be accessed over the PCIe bus. A different quote mentions being able to choose either DDR3 or GDDR5. I wasn't sure if that meant you can mix the two. I could envision a scenario in which you'd buy, for example, a motherboard with 2GB of GDDR5 soldered to it and then add extra DDR3 like normal. The amount of GDDR5 onboard would make for a nice product differentiator, which the motherboard manufacturers would probably like.
http://www.tweaktown.com/news/30038/amd-introduces-heterogeneous-uniform-memory-access/index.html
This morning AMD announced the next big advancement concerning their APU technology. AMD heterogeneous Uniform Memory Access (hUMA) is an intelligent computing architecture that enables the CPU, GPU, and other processors to work in harmony from a single piece of silicon in a single pool of memory and seamlessly move task to the best suited processing unit.
Mods please delete if repost.
Yes and no. The greatest benefits will be for IGPs (APUs), but the unified address space is coming to video cards, too. As to what works on video cards, as well, nVidia has the same thing coming along.This is huge news and I'm glad to see it. but will this only work for integrated GPUs?
Yes, the PlayStation 4 will have full implementation of HSA. It will be AMD's first heterogeneous based system on the market.is this happening in the PS4?
This is far from a architecture change. HSA is a huge stick that AMD is about to wap Intel with. If the software community takes in HSAIL (its already being looked over via C++ extensions), than a measly $150 A10-7800k will over throw the $1,000 i7 3960x in number crunching. Making even the best Intel has to offer at a enthusiast level look like garbage. You know what the on-die iGPU is capable of? The one in the Kaveri A10 will consist of GCN cores and AMD is talking 1 teraflop. This is a master plan AMD had up their sleeve ever since they bought ATI. They can't out perform Intel on the x86 level, but they can still product APU's that can out perform them. It all just hasn't unraveled until now. To give you and anyone else a basic understanding of how HSA works. Basically when you allocate memory to your application, it can be accessed by both the CPU and GPU cores. So when writing an application, with what is possible for a GPU to crunch. You could pass that off onto the iGPU to be crunched ten times faster than the CPU could. The difference is like writing a program in C#, and then re-writing the entire program in assembly. The performance change of desktop software will be that life changing without having to resort to inline code.More powerpoint slides about HSA.
Must be so great. The last time i read so much about something from AMD was before they launched Bulldozer.
Indeed it is, I can't wait to see if the software world eats up HSAIL. If it does we could be talking about some seriously number crunching on a cheap little processor. Even the cheapest low end dual core A4's will compete with the i5's and i7's if so. AMD has a shot to win the crown back, and for a very long time also (Intel iGPU's are crap, always have been).[ insert shut up and take my money pic ]
I can't wait, this kaveri stuff sounds interesting
Indeed but AMD's approach is much better, and if AMD choses to close up HSAIL (sorta like Nvidia did CUDA) than you can kiss HSA on Nvidia based products goodbye. I honestly think AMD should, simply because how Nvidia plays on the market. All of the good open source alternative came from AMD (OpenCL). HSA could make or break AMD. What I mean by that is HSA is such a technology if adopted, AMD will be back on top of the mountain. If not, than they will remain where they are at. Making good budget CPU's at a extremely cheap price. Tho as a programmer myself, I can tell you HSA is going to open a lot of doors.Yes and no. The greatest benefits will be for IGPs (APUs), but the unified address space is coming to video cards, too. As to what works on video cards, as well, nVidia has the same thing coming along.
This is far from a architecture change. HSA is a huge stick that AMD is about to wap Intel with. If the software community takes in HSAIL (its already being looked over via C++ extensions), than a measly $150 A10-7800k will over throw the $1,000 i7 3960x in number crunching.
Making even the best Intel has to offer at a enthusiast level look like garbage. You know what the on-die iGPU is capable of? The one in the Kaveri A10 will consist of GCN cores and AMD is talking 1 teraflop. This is a master plan AMD had up their sleeve ever since they bought ATI. They can't out perform Intel on the x86 level, but they can still product APU's that can out perform them. It all just hasn't unraveled until now.
To give you and anyone else a basic understanding of how HSA works. Basically when you allocate memory to your application, it can be accessed by both the CPU and GPU cores. So when writing an application, with what is possible for a GPU to crunch. You could pass that off onto the iGPU to be crunched ten times faster than the CPU could. The difference is like writing a program in C#, and then re-writing the entire program in assembly. The performance change of desktop software will be that life changing without having to resort to inline code.
All of the good open source alternative came from AMD (OpenCL). HSA could make or break AMD. What I mean by that is HSA is such a technology if adopted, AMD will be back on top of the mountain. If not, than they will remain where they are at. Making good budget CPU's at a extremely cheap price. Tho as a programmer myself, I can tell you HSA is going to open a lot of doors.
They can't out perform Intel on the x86 level
Sadly, no. They are moderately competitive in multithreaded applications (an FX-6300 or 8350, FI, is a better modern gaming CPU for the money than it is as a CAD CPU), but by sacrificing IPC, they can't make those gains turn into being faster, overall. Even with multithreaded applications, it takes particular cherry-picked ones to find cases where AMD bests Intel.They can in multi-threaded applications.
Lets start off by cutting down your rediculous post in size. OpenCL is not the same as HSA, so get that out of your head. Intel is not and wont be HSA compatible. The only way you will see HSA on a Intel based system is if 1) AMD open sources HSAIL, and 2) if Nvidia follows through with their plans of adding ARM cores to their GTX 700 series cards (I find it unlikely). So there is huge advantage for AMD, as only AMD based APU's will be capable of this new technology. I understand you know nothing about HSA nor OpenCL based on this single line of your post alone.Intel is full OpenCL compartible. So there is no advantage for AMD.
Intel is only improving upon graphics performance (e.g. game performance). Intel's graphics cores are terribly designed, and the architecture itself is far from mature. Also to let you know, you couldn't achieve 1 teraflop on any desktop chip Intel offers even when nitrogen cooled. So again, clearly you have no clue what you're talking about. To give you an idea of how wrong you are, a 3960x can't break 200 gflops when overclocked to its limit.I guess you don't know: Intel is improving their iGPU, too. :sneaky:
And 1 TFLOPs is nothing for 2014.
Both need their own share of work? This is exactly why HSA was created! When you're doing desktop work, your GPU is sitting there doing nothing. Graphics cores can crunch numbers over 10x faster than x86 cores. And no you cannot utilize HSA right now (it's not even fully implemented into their APU's for darned sake.). It works similar to OpenCL basing how it hands off a specific takes to the GPU to compute. But with unified memory, the CPU and GPU cores become one steam factory (laymans terms).That will not work. There is a reason why we have CPUs and GPUs. Both need their own share of work.
BTW: You can buy a OpenACC compile and do it with Cuda (C++) or OpenCL automatically right now if you want.
I don't believe in the hype, as a programmer I know exactly what HSA will bring to the table. And how it can be effectively utilized like I explained in my previous post. There's nothing hyped here except your butthurt attitude. You seem pretty defensive when someone presents real knowledge. :sneaky:Believe in the hype. D:
I wasn't referring to how heavily threaded an application is (as most are not and don't need to be). Simply put Intel is king of IPC right now on the x86 market. Tho yes if an application is heavily thread, they can out perform Intel's offerings (thus why people buy 8350's instead of i7's for video/3d work).They can in multi-threaded applications.
Sadly, no. They are moderately competitive in multithreaded applications (an FX-6300 or 8350, FI, is a better modern gaming CPU for the money than it is as a CAD CPU), but by sacrificing IPC, they can't make those gains turn into being faster, overall. Even with multithreaded applications, it takes particular cherry-picked ones to find cases where AMD bests Intel.
I wasn't referring to how heavily threaded an application is (as most are not and don't need to be). Simply put Intel is king of IPC right now on the x86 market. Tho yes if an application is heavily thread, they can out perform Intel's offerings (thus why people buy 8350's instead of i7's for video/3d work).
Sure, it's not. But that's the thing: you have to look. The vast majority of the time, Intel's got them beat.It is not difficult to find recent multithreaded soft where FX chips are faster than Intel chips.
Show me some benchmarks that show this great gain for HSA. Oh right, you cant, because neither the hardware or software exists yet.
You can insist that whatever you want is going to happen, but until there are independent benchmarks, it is only speculation.
:yawn: The classic AMD tasteless powerpoint slides, promising the moon and the stars, while making a smelly turd. If history tought us anything, I see another failboat afloat. Wake me up when I can buy it and get advantage in most of my apps.