Overclock.net › Forums › AMD › AMD CPUs › AMD No longer a viable option for mid-high end?
New Posts  All Forums:Forum Nav:

AMD No longer a viable option for mid-high end? - Page 86  

post #851 of 1593
Quote:
Originally Posted by AMDATI View Post

Many times better.....ONLY when compared to previous APU's. It really can't get all that much better either, since SYSRAM bandwidth limitations already cripple HSA APU's. They've hit a wall until they add onboard high bandwidth RAM, and that comes with its own logistical problems, meaning it's not happening anytime soon.

You completely miss the point here.

Every software and API written up until now is made from ground up with two very specific conditions in mind.

1. The GPU and CPU processors do not share the same resources.

2. The GPU has high bandwidth, slow latency integrated memory and its data has to go through the CPU and the SYSRAM if it can't go to the integrated memory.


Every API currently out there takes those 2 things as granted, and naturally, all software using every API out there does the same.
You can't use a feature if it's not exposed to you and you have no way to expose it. It's like having a 64bit processor with 16GB SYSRAM running Windows XP 32bit. The RAM is there, you can use some of it, but you can never use it from one point onwards unless you change OS.
post #852 of 1593
Quote:
Originally Posted by AMDATI View Post

Quote:
Originally Posted by sdlvx View Post

Yeah, but you know what? That's coming after a Maxwell Refresh. Maxwell refresh isn't due until late this year. Meaning we won't see Pascal until at least late 2015. Nvndia always gets up and starts talking about their future technology earlier than anyone else. Then, they make these ridiculous claims comparing the new thing to what is out already, and then imply that their new thing will be competing with the old ones. Or in cases of Tegra, they release a few months earlier than the competition, compare Tegra to the old chips from the competitor, wave their hands around talking about how amazing it is compared to year old hardware, and then competitors release updates which destroy Tegra and Tegra ends up not shipping in any serious devices.

As for NVLink, Nvidia is in the worst position out of Intel, AMD, and Nvidia to release a technology like this.

You are all right, PCIe is a massive hurdle for HSA and other shared memory technologies. It doesn't help that PCIe uses something similar to bits/second to measure speed while most of the other system is measured in bytes/second. Which means PCIe 4.0 is still only going to have 32GB/s of bandwidth. That sounds great compared to what we have, but by the time we see PCIe 4, we'll have DDR4. DDR4 is looking at 25GB/s per channel. Which means high end quad channel platforms will have 100GB/s bandwidth.

https://en.wikipedia.org/wiki/List_of_device_bandwidths

Note where HyperTransport stands. I think AMD is in the best position to replace PCIe with something far better. They did at one point have a HyperTransport Slot which was 100% pin compatible with PCIe. NVLink, as mentioned earlier, has no potential. At most it will end up a niche item in Power servers, which is far out of reach of probably anyone here. ANd if you do go that route, you can't really use Windows or Windows applications, so you end up with a Linux desktop best case scenario.

How cost effective do you think 100GB/s quad channel DDR4 will be? By the time all is said and done, you'll most likely end up paying more than you would have on a standalone GPU, with less performance overall to boot. let's also keep in mind that the bandwidth is still effectively halved as well, since if both the GPU and CPU are accessing the memory at the same time, regardless of if it's unified, that's going to eat into memory bandwidth.

and PCI-3 is still quite capable of providing enough bandwidth for even the highest end cards, which means PCI-3 isn't really a limiting factor to extreme performance.

HSA is strictly a performance improvement over previous APU system....nothing more. It cannot be compared to standalone GPU's at all, nor reasoned as some type of performance increase over standalone GPU's.

To put it simply, APU's, even for the next 5 years or more, will be nowhere near competing with standalone cards in both price and performance.

It depends on what the product is. For a high end setup where cost is not as important, it will be fine. My point is that APUs can get a lot more bandwidth then they currently have in an effort to address the fact that you're so hung up on APUs being bandwidth starved.

However, you're forgetting the key concept that "APUs are bandwidth starved" applies primarily to using the iGPU to play games when compared to dGPU performance with GDDR.

So, the point I'm making, again, is that APUs can scale with lots of bandwidth or they can remain with 2 channels or whatever.

But I don't think you understand the concept of HSA. iGPU gaming performance has absolutely nothing to do with HSA. And I do believe that this is where you are getting rather confused.

Between CPU, GPU, and system memory, PCIe 3.0 x16 is by far the slowest part. PCIe speed is measured in something equivalent to bits/second while memory uses bytes/second. I mentioned this earlier and clearly it fell on deaf ears, but PCIe 3.0 x16 only has about 16GB/s of bandwidth total. Not really comparable to a dual channel DDR3 setup running DDR3 1600. Yes, you don't even need to spend a ton of money on ram to get a ton of bandwidth.

I feel sort of bad picking on you since you don't even understand what HSA is, even though you're so vigorously attacking it, but if APUs are horribly memory bandwidth bottlenecked and can't compete with dGPUs, how do you explain this?



I think you are failing to get the point that there are just things that GPUs simply are unable to do, and the purpose of HSA is to be able to use the GPU in situations where you normally couldn't, or wouldn't want to because doing so would mean you'd have to move the entire application from system memory into video memory for GPGPU, then back into system memory to use it.

And you do you recall the part earlier about PCIe 3.0 being the slowest? That's what you have to do to move it all.
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
Tyrant Kuma
(13 items)
 
Starscythe
(13 items)
 
Mobility
(6 items)
 
CPUCPUCPUCPU
Opteron 8431 Opteron 8431 Opteron 8431 Opteron 8431 
MotherboardGraphicsRAMHard Drive
Super Micro h8qme-2+ Sapphire 4870 Samsung DDR2 ECC 667mhz western digital caviar blue 
CoolingCoolingOSPower
FX 8350 stock cooler FX 6300 stock cooler Gentoo Linux 2x Silverstone PSU 
Case
Custom Fabbed Steel Case 
CPUMotherboardGraphicsRAM
AMD FX 8350 gigabyte 990FXA-UD5 7970 Mushkin Enhanced Blackline 16GB 
OSOSMonitorKeyboard
Gentoo Linux Windows 7 x64 Yamakasi Catleap Q270 Corsair K90 
PowerCaseMouseAudio
Silverstone Strider Gold Evolution 850W 80 PLUS... Antec 1200 Logitech G9x Asus Xonar D2X 
Other
XSPC RS360 Raystorm with custom radiator mounts 
CPUGraphicsRAMOS
a4 5000 Radeon HD 8330 8GB G. Skill DDR3 1600 9-9-9 Gentoo Linux 
OSOther
Windows 7 Lenovo Thinkpad x140e 
  hide details  
post #853 of 1593
Thread Starter 
Quote:
Originally Posted by sdlvx View Post

It depends on what the product is. For a high end setup where cost is not as important, it will be fine. My point is that APUs can get a lot more bandwidth then they currently have in an effort to address the fact that you're so hung up on APUs being bandwidth starved.

However, you're forgetting the key concept that "APUs are bandwidth starved" applies primarily to using the iGPU to play games when compared to dGPU performance with GDDR.

So, the point I'm making, again, is that APUs can scale with lots of bandwidth or they can remain with 2 channels or whatever.

But I don't think you understand the concept of HSA. iGPU gaming performance has absolutely nothing to do with HSA. And I do believe that this is where you are getting rather confused.

Between CPU, GPU, and system memory, PCIe 3.0 x16 is by far the slowest part. PCIe speed is measured in something equivalent to bits/second while memory uses bytes/second. I mentioned this earlier and clearly it fell on deaf ears, but PCIe 3.0 x16 only has about 16GB/s of bandwidth total. Not really comparable to a dual channel DDR3 setup running DDR3 1600. Yes, you don't even need to spend a ton of money on ram to get a ton of bandwidth.

I feel sort of bad picking on you since you don't even understand what HSA is, even though you're so vigorously attacking it, but if APUs are horribly memory bandwidth bottlenecked and can't compete with dGPUs, how do you explain this?



I think you are failing to get the point that there are just things that GPUs simply are unable to do, and the purpose of HSA is to be able to use the GPU in situations where you normally couldn't, or wouldn't want to because doing so would mean you'd have to move the entire application from system memory into video memory for GPGPU, then back into system memory to use it.

And you do you recall the part earlier about PCIe 3.0 being the slowest? That's what you have to do to move it all.

I have an extensive grasp of the issue, and you're interpreting it incorrectly.

Now the obligatory thing to point out: People always try to point to something else. If it's a non gaming discussion, people cite gaming in hopes to get their inaccurate points across, and vice versa. So obviously it must be addressed on both fronts.

I understand all of how HSA works. The primary thing being shared memory space.

When you talk about "and the purpose of HSA is to be able to use the GPU in situations where you normally couldn't" that's simply not true. You can do everything with a standalone GPU, that you can with an APU.....and better at the same price point or less too.

HSA just includes some functions programs can call on so that they can use the GPU inside of the APU without having to program specifically for an APU environment (example: programs don't have to code specifically for how memory will be handled in an APU system, that's all done automatically). When it comes to HSA, its benefits only really apply against APU's without HSA. HSA however, does not do things you can't already do on any modern GPU. It doesn't magically process x86 code on the GPU for example. So, no it's not doing anything a GPU can't already do.

And again, PCIE doesn't need to be that fast. It's not really a bottleneck, even a $1,000 GPU can safely run on PICE3 with a huge amount of performance. I can't make this any more clear, the PICE bandwidth requirements really aren't all that much in the grand scheme of things. In reality, you're probably not even maxing your SYSRAM bandwidth in a CPU/GPU system when either needs to manipulate something in the others RAM, let alone your PCIE bandwidth.

GPU's are all about massive amounts of I/O. Input, Output. And it needs fast memory to be able to move not just large amounts of data, but also rapidly. Break up say a 100GB/s bus, and you get about 10GB every tenth of a second. It's also very much like how SSD's have high IOPs, even the first generation of SSD's had very bad read/write performance, but the high IOPS made a difference. It's less about what's happening in a whole second, and more about what's happening in less than the tenth of a second. For example, in order to get 60fps, your GPU has to be able to process each frame in 1/60th of a second. That requires enough bandwidth to be available in that time frame to achieve that (assuming the GPU power is up to snuff). Now keep in mind, when I talk about a frame, I'm not talking about the data that makes up the picture sent to your monitor.

Now let's think about it....How much bandwidth does a 20GB/s SYSRAM with APU have in 1/60th of a second? A mere 340MB. That's not much in the grand scheme of things, and 20GB/s is on the high end of SYSRAM speeds. Most common is 10GB/s, which brings it down to just 170MB. Obviously that's not enough I/O to draw a single frame in that amount of time. But there's also another downside, since RAM is better for for larger chunks over a longer period rather than tons of small chunks over a shorter period, so your transfer rate in those short time scales could actually be worse and uneven in reality. i.e. you might get more or less bandwidth in the same successive fractions of a second. The more you have in a given fraction of a second, the better. But if you have less than you need to work out that frame, then you have to eat into the bandwidth from the next fraction of a second to get enough data in or out of memory.

Now, How much bandwidth does a 150GB/s GDDR5 have in 1/60th of a second? A whopping 2.5GB, nearly 8 times more bandwidth in that given time frame of 1/60th of a second. That's clearly a lot more data to work with.



So if you don't have enough bandwidth in that fraction of time, you aren't going to have enough I/O to input data or output a frame, so your GPU just ends up waiting on Input or Output before it can do anything else, which on GPU processing speed timescale, is an eternity. It doesn't matter how powerful a GPU is, once it's limited on RAM bandwidth, it's not going to go any faster than the bandwidth speed allows it to change out input/output data. The GPU is effectively starved of data to do anything with.

This same thing also applies to non gaming GPU usage, since again, the GPU relies heavily on massive amounts of I/O. The GPU takes in some data, processes it, outputs it back to memory. The faster it gets the data outputted, the faster it can get new data inputted, and so on and so forth. This is a bedrock fact that simply can not be ignored nor downplayed, regardless of the usage scenario.
Edited by AMDATI - 3/26/14 at 2:40am
Not Yours
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k ASRock Z97E-ITX/AC MSI GTX 1080Ti G.SKILL DDR3  
Hard DriveHard DriveHard DriveCooling
SAMSUNG 830 OCZ Deneva 2R WD Scorpio Noctua U14S 
OSMonitorKeyboardPower
Windows 10 Acer XB270HU IPS CM Quickfire Rapid Fractal Design 650w 
CaseMouseMouse PadAudio
Node 304 SteelSeries Rival 700 SteelSeries Qck Sennheiser HD598 
  hide details  
Not Yours
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k ASRock Z97E-ITX/AC MSI GTX 1080Ti G.SKILL DDR3  
Hard DriveHard DriveHard DriveCooling
SAMSUNG 830 OCZ Deneva 2R WD Scorpio Noctua U14S 
OSMonitorKeyboardPower
Windows 10 Acer XB270HU IPS CM Quickfire Rapid Fractal Design 650w 
CaseMouseMouse PadAudio
Node 304 SteelSeries Rival 700 SteelSeries Qck Sennheiser HD598 
  hide details  
post #854 of 1593
Quote:
Originally Posted by AMDATI View Post

I have an extensive grasp of the issue, and you're interpreting it incorrectly.

Now the obligatory thing to point out: People always try to point to something else. If it's a non gaming discussion, people cite gaming in hopes to get their inaccurate points across, and vice versa. So obviously it must be addressed on both fronts.

I understand all of how HSA works. The primary thing being shared memory space.

When you talk about "and the purpose of HSA is to be able to use the GPU in situations where you normally couldn't" that's simply not true. You can do everything with a standalone GPU, that you can with an APU.....and better at the same price point or less too. HSA just includes some functions programs can call on so that they can use the GPU inside of the APU without having to program specifically for an APU environment. When it comes to HSA, its benefits only really apply against APU's without HSA. HSA however, does not do things you can't already do on any modern GPU. It doesn't magically process x86 code on the GPU for example.

And again, PCIE doesn't need to be that fast. It's not really a bottleneck, even a $1,000 GPU can safely run on PICE3 with a huge amount of performance. I can't make this any more clear, the PICE bandwidth requirements really aren't all that much in the grand scheme of things.

GPU's are all about massive amounts of I/O. Input, Output. And it needs fast memory to be able to copy not just large amounts of data, but also rapidly. Break up say a 100GB/s bus, and you get about 10GB every tenth of a second. It's also very much like how SSD's have high IOPs, even the first generation of SSD's had very bad read/write performance, but the high IOPS made a difference. It's less about what's happening in a whole second, and more about what's happening in less than the tenth of a second. for example, in order to get 60fps, your GPU has to be able to process each frame in 1/60th of a second. That requires enough bandwidth to be available in that time frame to achieve that.

Now let's think about it....How much bandwidth does a 20GB/s SYSRAM with APU have in 1/60th of a second? A mere 340MB. That's not much in the grand scheme of things, and 20GB/s is on the high end of SYSRAM speeds. Most common is 10GB/s, which brings it down to just 120MB. Obviously that's not enough I/O to draw a single frame in that amount of time.

Now, How much bandwidth does a 150GB/s GDDR5 have in 1/60th of a second? A whopping 2.5GB, nearly 8 times more bandwidth in that given time frame of 1/60th of a second.
Why do you assume that HSA is primarily meant for CPU+dGPU over PCIe?? As of now it is only meant for iGPU using hUMA, though hUMA is capable of expanding HSA to dGPU but latency is an issue over there. And that x86 code argument is a standard mocking against HSA, you don't need to run x86 code on GPU in order to comply with HSA. You can think of GPU as taking up all the FP calculations, possibly vector calculations (not exactly, but it does acr like a co-processor) in a very parallelized manner which is far far quicker than any FPU/vector ISA on any desktop processor.
The high IOPS and memory BW is only when dGPU will get involved in HSA environment, otherwise system memory offers good enough HSA performance.
    
CPUMotherboardGraphicsRAM
AMD FX-8150 ASUS M5A97 R2.0 AMD Radeon HD 6670 Corsair  
Hard DriveOptical DriveCoolingOS
Western Digital Caviar Blue Samsung Super WriteMaster AMD FX-8150 Stock Air Cooler Windows 8 PRO x64 
MonitorKeyboardPowerCase
Lenovo 17 inches CRT Lenovo Cooler Master Thunder 500W Cooler Master Elite 311 
MouseMouse PadAudio
Lenovo Lenovo Lenovo 
  hide details  
    
CPUMotherboardGraphicsRAM
AMD FX-8150 ASUS M5A97 R2.0 AMD Radeon HD 6670 Corsair  
Hard DriveOptical DriveCoolingOS
Western Digital Caviar Blue Samsung Super WriteMaster AMD FX-8150 Stock Air Cooler Windows 8 PRO x64 
MonitorKeyboardPowerCase
Lenovo 17 inches CRT Lenovo Cooler Master Thunder 500W Cooler Master Elite 311 
MouseMouse PadAudio
Lenovo Lenovo Lenovo 
  hide details  
post #855 of 1593
Quote:
Originally Posted by AMDATI View Post

I have an extensive grasp of the issue, and you're interpreting it incorrectly.

Now the obligatory thing to point out: People always try to point to something else. If it's a non gaming discussion, people cite gaming in hopes to get their inaccurate points across, and vice versa. So obviously it must be addressed on both fronts.

I understand all of how HSA works. The primary thing being shared memory space.

When you talk about "and the purpose of HSA is to be able to use the GPU in situations where you normally couldn't" that's simply not true. You can do everything with a standalone GPU, that you can with an APU.....and better at the same price point or less too. HSA just includes some functions programs can call on so that they can use the GPU inside of the APU without having to program specifically for an APU environment (example: programs don't have to code specifically for how memory will be handled in an APU system, that's all done automatically). When it comes to HSA, its benefits only really apply against APU's without HSA. HSA however, does not do things you can't already do on any modern GPU. It doesn't magically process x86 code on the GPU for example. So, no it's not doing anything a GPU can't already do.

And again, PCIE doesn't need to be that fast. It's not really a bottleneck, even a $1,000 GPU can safely run on PICE3 with a huge amount of performance. I can't make this any more clear, the PICE bandwidth requirements really aren't all that much in the grand scheme of things.

GPU's are all about massive amounts of I/O. Input, Output. And it needs fast memory to be able to move not just large amounts of data, but also rapidly. Break up say a 100GB/s bus, and you get about 10GB every tenth of a second. It's also very much like how SSD's have high IOPs, even the first generation of SSD's had very bad read/write performance, but the high IOPS made a difference. It's less about what's happening in a whole second, and more about what's happening in less than the tenth of a second. For example, in order to get 60fps, your GPU has to be able to process each frame in 1/60th of a second. That requires enough bandwidth to be available in that time frame to achieve that (assuming the GPU power is up to snuff).

Now let's think about it....How much bandwidth does a 20GB/s SYSRAM with APU have in 1/60th of a second? A mere 340MB. That's not much in the grand scheme of things, and 20GB/s is on the high end of SYSRAM speeds. Most common is 10GB/s, which brings it down to just 170MB. Obviously that's not enough I/O to draw a single frame in that amount of time. But there's also another downside, since SYSRAM is made more for large chunks rather than tons of small chunks, so your transfer rate in those short time scales could actually be worse and uneven in reality. i.e. you might get more or less bandwidth in the same successive fractions of a second.

Now, How much bandwidth does a 150GB/s GDDR5 have in 1/60th of a second? A whopping 2.5GB, nearly 8 times more bandwidth in that given time frame of 1/60th of a second.

So if you don't have enough bandwidth in that fraction of time, you aren't going to have enough I/O to input data or output a frame, so your GPU just ends up waiting on Input or Output before it can do anything else, which on GPU processing speed timescale, is an eternity.

This same thing also applies to non gaming GPU usage, since again, the GPU relies heavily on massive amounts of I/O. The GPU takes in some data, processes it, outputs it back to memory. The faster it gets the data outputted, the faster it can get new data inputted, and so on and so forth. This is a bedrock fact that simply can not be ignored nor downplayed.

Seriously, you need to educate yourself on these matters. What you're uttering makes no sense whatsoever. Do you even know what a GPU is supposed to do? Do you know how it does it? The way you do all that (mostly unrelated) math makes it seem like you don't know how rendering happens.
post #856 of 1593
Quote:
Originally Posted by imran27 View Post

Why do you assume that HSA is primarily meant for CPU+dGPU over PCIe?? As of now it is only meant for iGPU using hUMA, though hUMA is capable of expanding HSA to dGPU but latency is an issue over there. And that x86 code argument is a standard mocking against HSA, you don't need to run x86 code on GPU in order to comply with HSA. You can think of GPU as taking up all the FP calculations, possibly vector calculations (not exactly, but it does acr like a co-processor) in a very parallelized manner which is far far quicker than any FPU/vector ISA on any desktop processor.
The high IOPS and memory BW is only when dGPU will get involved in HSA environment, otherwise system memory offers good enough HSA performance.
Quote:
Originally Posted by PsyM4n View Post

Seriously, you need to educate yourself on these matters. What you're uttering makes no sense whatsoever. Do you even know what a GPU is supposed to do? Do you know how it does it? The way you do all that (mostly unrelated) math makes it seem like you don't know how rendering happens.
Ok AMDATI do you need any one else to tell you what most have been telling you all along? There is a point when one must realize that they have hit the limit of their knowledge and then must relent to others with more knowledge. That point was long ago. There is more to the story with HSA than the only one aspect you keep trying to play. Every software, hardware, and pretty much anything has good and bad points. But the fact you cling to the one and only singularity is sad. I reserve myself in the discussion because, though like you state, GPU have traditionally needed Higher bandwidth APUs use of memory is gonna be somewhat different. Let me 'splain my thoughts on the matter.

First there is the latency. In the case of the APU it is HEAVILY stacked in its favor. DDR3 has far lower latency and couple that with not needing the PCI-e lanes you have decreased latency further. Second having HUMA ,a part of HSA, decreases Latency even more. You can start to see how these things add up and how using your math starts to have far less meaning.

My brother now has his A10-7850K. At first his Mobo was setup with 13.12 AMD drivers. It performed like a typical low end 4 core, but with his last few computers he wasn't aware that it was subpar. Now most know the Mantle betas are mostly geared at the APUs and not so much the GPUs as traditional betas are. So he loaded it up and guess what, that greyhound was released from its cage. That sucker is moving. Also most will have 2133 speed ram with these which further closes the gap in performance, makes a huge difference in these APUs over what we normally see in regular CPUs.

I try not to get too bogged down in details with the APUs simply because they are a new beast and I am not the most versed in their inner workings or how the drivers AMD has come out with are leveraging their use. Judging by others responses to you MAYBE you should do the same.
post #857 of 1593
Thread Starter 
Quote:
Originally Posted by Durquavian View Post


Ok AMDATI do you need any one else to tell you what most have been telling you all along? There is a point when one must realize that they have hit the limit of their knowledge and then must relent to others with more knowledge. That point was long ago. There is more to the story with HSA than the only one aspect you keep trying to play. Every software, hardware, and pretty much anything has good and bad points. But the fact you cling to the one and only singularity is sad. I reserve myself in the discussion because, though like you state, GPU have traditionally needed Higher bandwidth APUs use of memory is gonna be somewhat different. Let me 'splain my thoughts on the matter.

First there is the latency. In the case of the APU it is HEAVILY stacked in its favor. DDR3 has far lower latency and couple that with not needing the PCI-e lanes you have decreased latency further. Second having HUMA ,a part of HSA, decreases Latency even more. You can start to see how these things add up and how using your math starts to have far less meaning.

My brother now has his A10-7850K. At first his Mobo was setup with 13.12 AMD drivers. It performed like a typical low end 4 core, but with his last few computers he wasn't aware that it was subpar. Now most know the Mantle betas are mostly geared at the APUs and not so much the GPUs as traditional betas are. So he loaded it up and guess what, that greyhound was released from its cage. That sucker is moving. Also most will have 2133 speed ram with these which further closes the gap in performance, makes a huge difference in these APUs over what we normally see in regular CPUs.

I try not to get too bogged down in details with the APUs simply because they are a new beast and I am not the most versed in their inner workings or how the drivers AMD has come out with are leveraging their use. Judging by others responses to you MAYBE you should do the same.

HSA latency improvements are only comparable to non HSA APU's. And the irony is, the latency isn't really all that substantial....at least not against standalone CPU/GPU's. Definitely not enough to offset lack of bandwidth compared to them. On traditional CPU/GPU systems, we're talking nanoseconds of latency here, not milliseconds.

I find it odd that you would try to educate me on something that you yourself admit you're not even versed in.

It's hard to make this any clearer. The benefits of newer APU's only apply against older APU's. This is why you primarily see benchmarking sites only comparing APU's to other APU's, and not to other GPU's.

There's a lot of people on here that clearly think they know more about computer tech than they actually do.

You have to use analogies like "greyhound released from its cage" because you don't actually understand the specs and details, so you use things like that to fill your gaps in knowledge.


Let's take the 5770 for example.....a 5 year old videocard, at release about $160 and its cheapest, was about $60 retail, and can be had for even less today used. It was a lower mid range card on release and was just a better energy version of the 4770 with DX11, a much older card, ancient by today's standards, and was even bested by the 4890.

Now let's compare the performance of that to the A10-7850K in Skyrim. Here's some benchmarks for both.

http://www.tomshardware.com/reviews/a10-7850k-a8-7600-kaveri,3725-7.html
http://www.tomshardware.com/reviews/skyrim-performance-benchmark,3074-4.html

1920x1080
Medium quality settings

5770: 75-84 FPS
7850K: 33-38 FPS

And skyrim is already a 3 year old game.

The 7850k can barely muster half the FPS of a 5 year old videocard that was barely mid range on its release. All for the 'low low' price of nearly $200 today! plus a bit more because hey, you need FM2+.....and maybe a bit more because hey, you may want pc2400 or higher. I mean the more you build an APU system from the ground up, the less cost effective it gets. So here you are, spending well over $300-400 on a system that gets less than half the performance of a 5 year old videocard that you can probably buy for $20 today.

So much for latency.
Edited by AMDATI - 3/26/14 at 3:32pm
Not Yours
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k ASRock Z97E-ITX/AC MSI GTX 1080Ti G.SKILL DDR3  
Hard DriveHard DriveHard DriveCooling
SAMSUNG 830 OCZ Deneva 2R WD Scorpio Noctua U14S 
OSMonitorKeyboardPower
Windows 10 Acer XB270HU IPS CM Quickfire Rapid Fractal Design 650w 
CaseMouseMouse PadAudio
Node 304 SteelSeries Rival 700 SteelSeries Qck Sennheiser HD598 
  hide details  
Not Yours
(16 items)
 
  
CPUMotherboardGraphicsRAM
i7 4790k ASRock Z97E-ITX/AC MSI GTX 1080Ti G.SKILL DDR3  
Hard DriveHard DriveHard DriveCooling
SAMSUNG 830 OCZ Deneva 2R WD Scorpio Noctua U14S 
OSMonitorKeyboardPower
Windows 10 Acer XB270HU IPS CM Quickfire Rapid Fractal Design 650w 
CaseMouseMouse PadAudio
Node 304 SteelSeries Rival 700 SteelSeries Qck Sennheiser HD598 
  hide details  
post #858 of 1593
Quote:
Originally Posted by AMDATI View Post

HSA latency improvements are only comparable to non HSA APU's. And the irony is, the latency isn't really all that substantial....at least not against standalone CPU/GPU's. Definitely not enough to offset lack of bandwidth compared to them. On traditional CPU/GPU systems, we're talking nanoseconds of latency here, not milliseconds.

I find it odd that you would try to educate me on something that you yourself admit you're not even versed in.

It's hard to make this any clearer. The benefits of newer APU's only apply against older APU's. This is why you primarily see benchmarking sites only comparing APU's to other APU's, and not to other GPU's.

There's a lot of people on here that clearly think they know more about computer tech than they actually do.

You have to use analogies like "greyhound released from its cage" because you don't actually understand the specs and details, so you use things like that to fill your gaps in knowledge.


Let's take the 5770 for example.....a 5 year old videocard, at release about $160 and its cheapest, was about $60 retail, and can be had for even less today used. It was a lower mid range card on release and was just a better energy version of the 4770 with DX11, a much older card, ancient by today's standards, and was even bested by the 4890.

Now let's compare the performance of that to the A10-7850K in Skyrim. Here's some benchmarks for both.

http://www.tomshardware.com/reviews/a10-7850k-a8-7600-kaveri,3725-7.html
http://www.tomshardware.com/reviews/skyrim-performance-benchmark,3074-4.html

1920x1080
Medium quality settings

5770: 75-84 FPS
7850K: 33-38 FPS

And skyrim is already a 3 year old game.

The 7850k can barely muster half the FPS of a 5 year old videocard that was barely mid range on its release. All for the 'low low' price of nearly $200 today! plus a bit more because hey, you need FM2+.....and maybe a bit more because hey, you may want pc2400 or higher. I mean the more you build an APU system from the ground up, the less cost effective it gets. So here you are, spending well over $300-400 on a system that gets less than half the performance of a 5 year old videocard that you can probably buy for $20 today.
Medium settings require more memory speed to update frames faster. It would actually be better testing at higher settings. Anyway, I'm not in the mood arguing with someone comparing cards by a directx 9 game from times long ago.
The Machine
(14 items)
 
Nexus 7 2013
(11 items)
 
 
CPUMotherboardGraphicsRAM
A10 6800K Asus F2A85-V MSI 6870 Hawx, VTX3D 5770, AMD HD6950(RIP), Sap... G.skill Ripjaws PC12800 6-8-6-24 
Hard DriveOptical DriveOSMonitor
Seagate 7200.5 1TB NEC 3540 Dvd-Rom Windows 7 x32 Ultimate Samsung P2350 23" 1080p 
PowerCaseMouseAudio
Seasonic s12-600w CoolerMaster Centurion 5 Logitech G600 Auzen X-Fi Raider 
CPUMotherboardGraphicsRAM
Quad Krait 300 at 1.5Ghz Qualcomm APQ8064-1AA SOC Adreno 320 at 400mhz 2GB DDR3L-1600 
Hard DriveOSMonitorKeyboard
32GB Internal NAND Android 5.0 7" 1920X1200 103% sRGB & 572 cd/m2 LTPS IPS Microsoft Wedge Mobile Keyboard 
PowerAudio
3950mAh/15.01mAh Battery Stereo Speakers 
  hide details  
The Machine
(14 items)
 
Nexus 7 2013
(11 items)
 
 
CPUMotherboardGraphicsRAM
A10 6800K Asus F2A85-V MSI 6870 Hawx, VTX3D 5770, AMD HD6950(RIP), Sap... G.skill Ripjaws PC12800 6-8-6-24 
Hard DriveOptical DriveOSMonitor
Seagate 7200.5 1TB NEC 3540 Dvd-Rom Windows 7 x32 Ultimate Samsung P2350 23" 1080p 
PowerCaseMouseAudio
Seasonic s12-600w CoolerMaster Centurion 5 Logitech G600 Auzen X-Fi Raider 
CPUMotherboardGraphicsRAM
Quad Krait 300 at 1.5Ghz Qualcomm APQ8064-1AA SOC Adreno 320 at 400mhz 2GB DDR3L-1600 
Hard DriveOSMonitorKeyboard
32GB Internal NAND Android 5.0 7" 1920X1200 103% sRGB & 572 cd/m2 LTPS IPS Microsoft Wedge Mobile Keyboard 
PowerAudio
3950mAh/15.01mAh Battery Stereo Speakers 
  hide details  
post #859 of 1593
ALL OF THOSE COMPLAINING! Listen HSA is only capable on processors that support it. There is a such thing as GPGPU tech but it's not at all like HSA because it needs to operate in junction with the CPU. dGPUs just don't have that pleasure yet. iGPUs on the Kaveri are not at all similar to the ones in the R7. It can handle many x86 threads as long as it's encoded to pick up the task. That's why you have Compute Cores. These GPU cores can do Prime and various other complicated mathematical operations yet also do graphics, it behaves more like a vector card where dGPU are more like raster cards. Almost everything we do on the computer today can be ran by the iGPU built on the Kaveri processors, again it just need modified instructions from the x86 platform to be threaded through the CU cores.
Edited by SpeedyVT - 3/26/14 at 4:35pm
Power Tower
(22 items)
 
SteamBox
(9 items)
 
Doge Miner
(7 items)
 
CPUMotherboardGraphicsRAM
Ryzen 1700X AX370-Gaming 5 AMD Radeon R9 200 Series G.Skill DDR4-2400 
RAMRAMRAMHard Drive
G.Skill DDR4-2400 G.Skill DDR4-2400 G.Skill DDR4-2400 Samsung 840 Pro 
Hard DriveHard DriveHard DriveHard Drive
CX300 Crucial 480GB Toshiba 4TB Toshbia 4TB Western Digital Black 1TB 
CoolingOSMonitorMonitor
h110i Windows 10 42" LG TV 20" Digitizer ASUS 
KeyboardPowerCaseMouse
Corsair Vengeance Mechanical Keyboard  850watt Vampire Gold Rated NZXT S340 Elite Corsair RGB FPS Mouse 
Mouse PadAudio
Borderlands Mousepad Realtek HD 
  hide details  
Power Tower
(22 items)
 
SteamBox
(9 items)
 
Doge Miner
(7 items)
 
CPUMotherboardGraphicsRAM
Ryzen 1700X AX370-Gaming 5 AMD Radeon R9 200 Series G.Skill DDR4-2400 
RAMRAMRAMHard Drive
G.Skill DDR4-2400 G.Skill DDR4-2400 G.Skill DDR4-2400 Samsung 840 Pro 
Hard DriveHard DriveHard DriveHard Drive
CX300 Crucial 480GB Toshiba 4TB Toshbia 4TB Western Digital Black 1TB 
CoolingOSMonitorMonitor
h110i Windows 10 42" LG TV 20" Digitizer ASUS 
KeyboardPowerCaseMouse
Corsair Vengeance Mechanical Keyboard  850watt Vampire Gold Rated NZXT S340 Elite Corsair RGB FPS Mouse 
Mouse PadAudio
Borderlands Mousepad Realtek HD 
  hide details  
post #860 of 1593
For AMDATI
Ok then I guess the gloves are off. Your inept attempts to prove your ignorance superior to anyone in here is laughable. It is obvious that you know next to nothing on the subject of HSA and the A10-7850K. Every point you make is proven wrong by a vast number of people in the thread . By all means, if you wish to see who knows more then lets have at it.

First lets look at your last post:
Quote:
HSA latency improvements are only comparable to non HSA APU's. And the irony is, the latency isn't really all that substantial....at least not against standalone CPU/GPU's. Definitely not enough to offset lack of bandwidth compared to them. On traditional CPU/GPU systems, we're talking nanoseconds of latency here, not milliseconds.
WRONG. Each part that adds latency and is existent in dGPU systems but not in APUs adds up to make a considerable difference. To disregard shows ignorance and leaves the door open WIDE for rebuttal.
Quote:
You have to use analogies like "greyhound released from its cage" because you don't actually understand the specs and details, so you use things like that to fill your gaps in knowledge.

Oh I understand quite intimately how it works and seeing how you have been corrected a lot by numerous people only shows your inability to see your own shortcoming in this topic. And the greyhounds part was an attempt to dumb it down enough for you to understand, but apparently still over your head. Too bad you cant see the "not shocked" look on my face.
Quote:
Now let's compare the performance of that to the A10-7850K in Skyrim. Here's some benchmarks for both.

http://www.tomshardware.com/reviews/a10-7850k-a8-7600-kaveri,3725-7.html
http://www.tomshardware.com/reviews/skyrim-performance-benchmark,3074-4.html

1920x1080
Medium quality settings

5770: 75-84 FPS
7850K: 33-38 FPS

And skyrim is already a 3 year old game.

The 7850k can barely muster half the FPS of a 5 year old videocard that was barely mid range on its release. All for the 'low low' price of nearly $200 today! plus a bit more because hey, you need FM2+.....and maybe a bit more because hey, you may want pc2400 or higher. I mean the more you build an APU system from the ground up, the less cost effective it gets. So here you are, spending well over $300-400 on a system that gets less than half the performance of a 5 year old videocard that you can probably buy for $20 today.
Ok I want to thank you for this opportunity and the 35 others before where , remember that door comment earlier, you left the door WIDE OPEN.

http://benchmarkreviews.com/13238/amd-a10-7850k-performance-optimized-catalyst-14-2-driver/

Its a good start to get you well informed. I would like to have found a 14.3beta test on the A10-7850k but alas didn't easily find it and wasting too much of my time on worthlessness well is a waste. Your review you posted is a preliminary meaning prior to release and well here is what we have now:



Amazing improvement. Oh but I forgot you think the 5770 means something here so lets splain some things.

First that review with the 5770 was with the i5-2500k, wait there is more. Skyrim uses X87 instruction set code and Intel owns that, wait there is more. AMD no longer supports X87 code in its chips since bulldozer, wait there is more. So one must use the Bulldozer Conditioner by Stilt ( thanx Stilt ), wait there is more. I just reinstalled Skyrim, actually reinstalled Windows7, so Skyrim reinstall was a must, wait there is more. Started Skyrim up with a ton of mods and a few that didn't play well with the others, which unfortunately caused lower than normal framerates, but for this point makes a very good example, and you guessed it: There is more. Was getting 25FPS before and 55FPS (vsync enabled) after using the conditioner. OMG that is amazing. Imagine what the 7850K might do.

Well that is one problem with your test example, the other being Skyrim is typically known to be a CPU bound game(at least till you mod it hardcore) and therefore the number one post for benches against AMD along with WoW and SC2. None of which show GPU prowess.

And for the finale and biggest problem for your argument ever: You can pair a dGPU with an APU. OMG that is amazing. Guess you didn't know that. And guess what else... Nevermind that's a long wait. The iGPU is still existent so it is still HSA able and Mantle can still leverage it as well if need be.

Ok I know that is a lot to take in so take a day or 2 and absorb it. And for the love of God, come prepared and understand more than 2 keywords if you wish to debate.
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: AMD CPUs
This thread is locked  
Overclock.net › Forums › AMD › AMD CPUs › AMD No longer a viable option for mid-high end?