AMD's Radeon HD 5870: Bringing About the Next Generation Of GPUs
by Ryan Smith on September 23, 2009 9:00 AM EST- Posted in
- GPUs
More GDDR5 Technologies: Memory Error Detection & Temperature Compensation
As we previously mentioned, for Cypress AMD’s memory controllers have implemented a greater part of the GDDR5 specification. Beyond gaining the ability to use GDDR5’s power saving abilities, AMD has also been working on implementing features to allow their cards to reach higher memory clock speeds. Chief among these is support for GDDR5’s error detection capabilities.
One of the biggest problems in using a high-speed memory device like GDDR5 is that it requires a bus that’s both fast and fairly wide - properties that generally run counter to each other in designing a device bus. A single GDDR5 memory chip on the 5870 needs to connect to a bus that’s 32 bits wide and runs at base speed of 1.2GHz, which requires a bus that can meeting exceedingly precise tolerances. Adding to the challenge is that for a card like the 5870 with a 256-bit total memory bus, eight of these buses will be required, leading to more noise from adjoining buses and less room to work in.
Because of the difficulty in building such a bus, the memory bus has become the weak point for video cards using GDDR5. The GPU’s memory controller can do more and the memory chips themselves can do more, but the bus can’t keep up.
To combat this, GDDR5 memory controllers can perform basic error detection on both reads and writes by implementing a CRC-8 hash function. With this feature enabled, for each 64-bit data burst an 8-bit cyclic redundancy check hash (CRC-8) is transmitted via a set of four dedicated EDC pins. This CRC is then used to check the contents of the data burst, to determine whether any errors were introduced into the data burst during transmission.
The specific CRC function used in GDDR5 can detect 1-bit and 2-bit errors with 100% accuracy, with that accuracy falling with additional erroneous bits. This is due to the fact that the CRC function used can generate collisions, which means that the CRC of an erroneous data burst could match the proper CRC in an unlikely situation. But as the odds decrease for additional errors, the vast majority of errors should be limited to 1-bit and 2-bit errors.
Should an error be found, the GDDR5 controller will request a retransmission of the faulty data burst, and it will keep doing this until the data burst finally goes through correctly. A retransmission request is also used to re-train the GDDR5 link (once again taking advantage of fast link re-training) to correct any potential link problems brought about by changing environmental conditions. Note that this does not involve changing the clock speed of the GDDR5 (i.e. it does not step down in speed); rather it’s merely reinitializing the link. If the errors are due the bus being outright unable to perfectly handle the requested clock speed, errors will continue to happen and be caught. Keep this in mind as it will be important when we get to overclocking.
Finally, we should also note that this error detection scheme is only for detecting bus errors. Errors in the GDDR5 memory modules or errors in the memory controller will not be detected, so it’s still possible to end up with bad data should either of those two devices malfunction. By the same token this is solely a detection scheme, so there are no error correction abilities. The only way to correct a transmission error is to keep trying until the bus gets it right.
Now in spite of the difficulties in building and operating such a high speed bus, error detection is not necessary for its operation. As AMD was quick to point out to us, cards still need to ship defect-free and not produce any errors. Or in other words, the error detection mechanism is a failsafe mechanism rather than a tool specifically to attain higher memory speeds. Memory supplier Qimonda’s own whitepaper on GDDR5 pitches error correction as a necessary precaution due to the increasing amount of code stored in graphics memory, where a failure can lead to a crash rather than just a bad pixel.
In any case, for normal use the ramifications of using GDDR5’s error detection capabilities should be non-existent. In practice, this is going to lead to more stable cards since memory bus errors have been eliminated, but we don’t know to what degree. The full use of the system to retransmit a data burst would itself be a catch-22 after all – it means an error has occurred when it shouldn’t have.
Like the changes to VRM monitoring, the significant ramifications of this will be felt with overclocking. Overclocking attempts that previously would push the bus too hard and lead to errors now will no longer do so, making higher overclocks possible. However this is a bit of an illusion as retransmissions reduce performance. The scenario laid out to us by AMD is that overclockers who have reached the limits of their card’s memory bus will now see the impact of this as a drop in performance due to retransmissions, rather than crashing or graphical corruption. This means assessing an overclock will require monitoring the performance of a card, along with continuing to look for traditional signs as those will still indicate problems in memory chips and the memory controller itself.
Ideally there would be a more absolute and expedient way to check for errors than looking at overall performance, but at this time AMD doesn’t have a way to deliver error notices. Maybe in the future they will?
Wrapping things up, we have previously discussed fast link re-training as a tool to allow AMD to clock down GDDR5 during idle periods, and as part of a failsafe method to be used with error detection. However it also serves as a tool to enable higher memory speeds through its use in temperature compensation.
Once again due to the high speeds of GDDR5, it’s more sensitive to memory chip temperatures than previous memory technologies were. Under normal circumstances this sensitivity would limit memory speeds, as temperature swings would change the performance of the memory chips enough to make it difficult to maintain a stable link with the memory controller. By monitoring the temperature of the chips and re-training the link when there are significant shifts in temperature, higher memory speeds are made possible by preventing link failures.
And while temperature compensation may not sound complex, that doesn’t mean it’s not important. As we have mentioned a few times now, the biggest bottleneck in memory performance is the bus. The memory chips can go faster; it’s the bus that can’t. So anything that can help maintain a link along these fragile buses becomes an important tool in achieving higher memory speeds.
Facebook
Twitter
RSS
327 Comments
View All Comments
ClownPuncher - Wednesday, September 23, 2009 - link
Absolutely, I can answer that for you.Those 2 "ports" you see are for aesthetic purposes only, the card has a shroud internally so those 2 ports neither intake nor exhaust any air, hot or otherwise.
Ryan Smith - Wednesday, September 23, 2009 - link
ClownPuncher gets a cookie. This is exactly correct; the actual fan shroud is sealed so that air only goes out the front of the card to go outside of the case. The holes do serve a cooling purpose though; allow airflow to help cool the bits of the card that aren't hooked up to the main cooler; various caps and what have you.SiliconDoc - Wednesday, September 23, 2009 - link
Ok good, now we know.So the problem now moves to the tiny 1/2 exhaust port on the back, did you stick your hand there and see how much that is blowing ? Does it whistle through there ? lol
Same amount of air(or a bit less) in half the exit space... that's going to strain the fan and or/reduce flow, no matter what anyone claims to the contrary.
It sure looks like ATI is doing a big favor to aftermarket cooler vendors.
GhandiInstinct - Wednesday, September 23, 2009 - link
Ryan,Developers arent pushing graphics anymore. Its not economnical, PC game supports is slowing down, everything is console now which is DX9. what purpose does this ATI serve with DX11 and all this other technology that won't even make use of games 2 years from now?
Waste of money..
ClownPuncher - Wednesday, September 23, 2009 - link
Clearly he should stop reviewing computer technology like this because people like you are content with gaming on their Wii and iPhone.This message has been brought to you by Sarcasm.
Griswold - Wednesday, September 23, 2009 - link
So you're echoing what nvidia recently said, when they claimed dx11/gaming on the PC isnt all that (anymore)? I guess nvidia can close shop (at least the gaming relevant part of it) now and focus on GPGPU. Why wait for GT300 as a gamer?Oh right, its gonna be blasting past the 5xxx and suddenly dx11 will be the holy grail again... I see how it is.
SiliconDoc - Wednesday, September 23, 2009 - link
rofl- It's great to see red roosters not crowing and hopping around flapping their wings and screaming nvidia is going down.Don't take any of this personal except the compliments, you're doing a fine job.
It's nice to see you doing my usual job, albiet from the other side, so allow me to compliment your fine perceptions. Sweltering smart.
But, now, let's not forget how ambient occlusion got poo-pooed here and shading in the game was said to be "an irritant" when Nvidia cards rendered it with just driver changes for the hardware. lol
Then of course we heard endless crowing about "tesselation" for ati.
Now it's what, SSAA (rebirthed), and Eyefinity, and we'll hear how great it is for some time to come. Let's not forget the endless screeching about how terrible and useless PhysX is by Nvidia, but boy when "open standards" finally gets "Havok and Ati" cranking away, wow the sky is the limit for in game destruction and water movement and shooting and bouncing, and on and on....
Of course it was "Nvidia's fault" that "open havok" didn't happen.
I'm wondering if 30" top resolution will now be "all there is!" for the next month or two until Nvidia comes out with their next generation - because that was quite a trick switching from top rez 30" DOWN to 1920x when Nvidia put out their 2560x GTX275 driver and it whomped Ati's card at 30" 2560x, but switched places at 1920x, which was then of course "the winning rez" since Ati was stuck there.
I could go on but you're probably fuming already and will just make an insult back so let the spam posting IZ2000 or whatever it's name will be this time handle it.
BTW there's a load of bias in the article and I'll be glad to point it out in another post, but the reason the red rooster rooting is not going beyond any sane notion of "truthful" or even truthiness, is because this 5870 Ati card is already percieved as " EPIC FAIL" !
I cannot imagine this is all Ati has, and if it is they are in deep trouble I believe.
I suspect some further releases with more power soon.
Finally - Wednesday, September 23, 2009 - link
Team Green - full foam ahead!*hands over towel*
There you go. Keep on foaming, I'm all amused :)
araczynski - Wednesday, September 23, 2009 - link
is DirectX11 going to be as worthless as 10? in terms of being used in any meaningful way in a meaningful amount of games?my 2 4850's are still keeping me very happy in my 'ancient' E8500.
curious to see how this compares to whatever nvidia rolls out, probably more of the same, better in some, worse in others, bottom line will be the price.... maybe in a year or two i'll build a new system.
of course by that time these'll be worthless too.
SiliconDoc - Wednesday, September 23, 2009 - link
Well it's certainly going to be less useful than PhysX, which is here said to be worthless, but of course DX11 won't get that kind of dissing, at least not for the next two months or so, before NVidia joins in.Since there's only 1 game "kinda ready" with DX11, I suppose all the hype and heady talk will have to wait until... until... uhh.. the 5870's are actually available and not just listed on the egg and tiger.
Here's something else in the article I found so very heartwarming:
---
" Wrapping things up, one of the last GPGPU projects AMD presented at their press event was a GPU implementation of Bullet Physics, an open source physics simulation library. Although they’ll never admit it, AMD is probably getting tired of being beaten over the head by NVIDIA and PhysX; Bullet Physics is AMD’s proof that they can do physics too. "
---
Unfortunately for this place,one of my friends pointed me to this little expose' that show ATI uses NVIDIA CARDS to develope "Bullet Physics" - ROFLMAO
-
" We have seen a presentation where Nvidia claims that Mr. Erwin Coumans, the creator of Bullet Physics Engine, said that he developed Bullet physics on Geforce cards. The bad thing for ATI is that they are betting on this open standard physics tech as the one that they want to accelerate on their GPUs.
"ATI’s Bullet GPU acceleration via Open CL will work with any compliant drivers, we use NVIDIA Geforce cards for our development and even use code from their OpenCL SDK, they are a great technology partner. “ said Erwin.
This means that Bullet physics is being developed on Nvidia Geforce cards even though ATI is supposed to get driver and hardware acceleration for Bullet Physics."
---
rofl - hahahahahha now that takes the cake!
http://www.fudzilla.com/content/view/15642/34/">http://www.fudzilla.com/content/view/15642/34/
--
Boy do we "hate PhysX" as ati fans, but then again... why not use the nvidia PhysX card to whip up some B Physics, folks I couldn't make this stuff up.