Adapting to Parallelism: Catalyst 5.12 More Dual Core Friendly?
by Derek Wilson on December 4, 2005 10:45 AM EST- Posted in
- GPUs
Performance Comparison: Cat 5.11 vs. Cat 5.12
The first thing we wanted to look at is the difference in scaling between driver versions. The following tables will show percent performance improvement of the beta 5.12 driver over that of the 5.11 driver. We will show performance improvements for both single and dual core configurations when moving to the beta driver.
While these numbers are, in fact, what users of single or dual core systems will experience when upgrading to newer drivers, there are other useful bits of information we can extract from them. We will be keeping an eye out for cases where the 5.12 driver performs worse than the 5.11 driver (these will be negative percentages in our tables). If, for instance, one tests shows the 5.12 driver doing worse in a single core platform and better in a dual core platform, we can discount some of the "value" of the dual core performance improvement as it's just making up for the performance hit on the single core side.
And as we can see from our Battlefield 2 test, The 5.11 driver performs as good as the 5.12 driver with no AA in 3 out of 6 tests. In the 8x6 case, the 5.11 driver handily bests the 5.12 beta. Enabling dual core allows the 5.12 driver to make up more than the ground it looses in single core performance, but the trade off just doesn't look good from this test.
And if you didn't think things could get worse, then just glance at the next table. The 5.12 driver tanks across the board on 4xAA performance under BF2. There isn't much more to say about this one.
Without AA, playing DoD:S, the 5.12 driver performs almost identically to the 5.11 driver on single core systems. Flipping the switch gives us an instant boost at 8x6 and 10x12, and even a little nudge in the right direction at 1600x1200.
Enabling 4xAA doesn't seem to change much. We see a little more benefit (percentage wise) when using 5.12 under dual core in 800x600 and 1600x1200, but the gain over 5.11 at 1024x768 drops a little. Either way, Day of Defeat Source seems to show that theres definitely a little benefit to be had by upgrading dual core systems to 5.12 from 5.11 drivers.
There are a few cases where the 5.12 driver improves performance in FarCry over the 5.11 even without the aide of dual core. Even though we see high percentage improvement with 5.12 under dual core, some of this could be general improvements to the way ATI handles the game.
Again, even with 4xAA FarCry benefits from the 5.12 drivers in 4 out of 6 tests (with both of those tests being much more GPU limited at 1600x1200).
There isn't much to say other than there isn't any improvement under Quake 4 when upgrading to the 5.12 drivers.
Which brings us to the test with the least change of all: Quake 4 with 4xAA.
Now let's take a look at performance improvement from a different perspective: improvement of a dual core system over a single core system.
The first thing we wanted to look at is the difference in scaling between driver versions. The following tables will show percent performance improvement of the beta 5.12 driver over that of the 5.11 driver. We will show performance improvements for both single and dual core configurations when moving to the beta driver.
While these numbers are, in fact, what users of single or dual core systems will experience when upgrading to newer drivers, there are other useful bits of information we can extract from them. We will be keeping an eye out for cases where the 5.12 driver performs worse than the 5.11 driver (these will be negative percentages in our tables). If, for instance, one tests shows the 5.12 driver doing worse in a single core platform and better in a dual core platform, we can discount some of the "value" of the dual core performance improvement as it's just making up for the performance hit on the single core side.
And as we can see from our Battlefield 2 test, The 5.11 driver performs as good as the 5.12 driver with no AA in 3 out of 6 tests. In the 8x6 case, the 5.11 driver handily bests the 5.12 beta. Enabling dual core allows the 5.12 driver to make up more than the ground it looses in single core performance, but the trade off just doesn't look good from this test.
| Battlefield 2 Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | -9.19 | -0.41 | 0.84 |
| Dual Core | 10.63 | 2.67 | -0.21 |
And if you didn't think things could get worse, then just glance at the next table. The 5.12 driver tanks across the board on 4xAA performance under BF2. There isn't much more to say about this one.
| Battlefield 2 4xAA Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | -5.05 | -2.87 | -0.89 |
| Dual Core | -4.42 | -0.15 | -1.17 |
Without AA, playing DoD:S, the 5.12 driver performs almost identically to the 5.11 driver on single core systems. Flipping the switch gives us an instant boost at 8x6 and 10x12, and even a little nudge in the right direction at 1600x1200.
| Day of Defeat Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 0.19 | 0.19 | 0.66 |
| Dual Core | 6.31 | 6.34 | 1.97 |
Enabling 4xAA doesn't seem to change much. We see a little more benefit (percentage wise) when using 5.12 under dual core in 800x600 and 1600x1200, but the gain over 5.11 at 1024x768 drops a little. Either way, Day of Defeat Source seems to show that theres definitely a little benefit to be had by upgrading dual core systems to 5.12 from 5.11 drivers.
| Day of Defeat 4xAA Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 0.37 | 0.19 | -0.27 |
| Dual Core | 6.69 | 4.31 | 2.66 |
There are a few cases where the 5.12 driver improves performance in FarCry over the 5.11 even without the aide of dual core. Even though we see high percentage improvement with 5.12 under dual core, some of this could be general improvements to the way ATI handles the game.
| FarCry Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 0.11 | 3 | -0.15 |
| Dual Core | 7.58 | 6.59 | 3.07 |
Again, even with 4xAA FarCry benefits from the 5.12 drivers in 4 out of 6 tests (with both of those tests being much more GPU limited at 1600x1200).
| FarCry 4xAA Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 2.8 | 1.51 | -1.5 |
| Dual Core | 7.26 | 4.92 | -0.75 |
There isn't much to say other than there isn't any improvement under Quake 4 when upgrading to the 5.12 drivers.
| Quake 4 Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 0.27 | 0.36 | 0 |
| Dual Core | 0.27 | 0.36 | 0.31 |
Which brings us to the test with the least change of all: Quake 4 with 4xAA.
| Quake 4 4xAA Percent Increase (Cat 5.11 to 5.12) | |||
| 800x600 | 1024x768 | 1600x1200 | |
| Single Core | 0.21 | 0.17 | 0 |
| Dual Core | -0.21 | 0 | 0 |
Now let's take a look at performance improvement from a different perspective: improvement of a dual core system over a single core system.
Facebook
Twitter
RSS
56 Comments
View All Comments
DrZoidberg - Wednesday, December 7, 2005 - link
lol totally ownedgreat link semiconductorslave
heres another link took 5 secs to find since almost all the websites show AMD is better
http://www.xbitlabs.com/articles/cpu/display/28cpu...">http://www.xbitlabs.com/articles/cpu/display/28cpu...
porkster - Tuesday, December 6, 2005 - link
You are failing to see that the majority of AMD CPU's are legacy as they have poor ability for modern software. Only AMD's top end CPU's are anything to consider, else all the others aren't suitable. AMD only perform well in single tasks. Intel are the KINGS of multitasking and bandwidth for the whole of their range.Games like DoD2 and Doom2 are not pushing the gfx routines and system, Black and White 2 does. B&w2 is a front line game, where the games in this test are safe bets for all processors. The point is this test was to measure performance for a dual-core device driver, so you would expect to see best of software.
SemiconductorSlave - Tuesday, December 6, 2005 - link
Well I'll agree that Hyperthreading benifits the P4 for multitasking, but lets not consider legacy, when comparing each companies latest dual cores, the Intel Pentium Processor Extreme Edition 840 $1029 at newegg, and the Athlon X2 4800+ $787 at neweggFrom Hardocp.com http://www.hardocp.com/article.html?art=NzY2">http://www.hardocp.com/article.html?art=NzY2
AMD uses what they call Direct Connect architecture. Instead of two processor cores being saddled to one bus and run to a single memory controller as we see with Intel dual core technology, we have to remember that AMD Athlon 64 processors have the memory controller on the CPU die itself and therefore no “front side bus” is needed. So each CPU on our dual core X2 has a much quicker route to the memory controller as with current Athlon 64 processors.So still the biggest benefit to the entire K8 core system is shining through in AMD's Athlon X2 line in the ways of HyperTransport and its extremely wide bus width when compared to Intel’s dual core 800MHz bus.
Sandra Memory Bandwith
Intel EE 840 Dual Core 4331
AMD X2 4800+ 5801
From same article
"On the dual core front, when you look at AMD’s flagship Athlon 64 X2 side by side with Intel’s flagship Pentium Extreme Edition 840 with HyperThreading, the obvious HyperThreading advantages seemingly disappear. In comparing single threaded applications, the Athlon 64 X2 4800+ shines over Intel’s 840 in our benchmarks."
porkster - Tuesday, December 6, 2005 - link
First off, sorry for spelling and typo's in last post. I reply on EDIT alot which this commenting system doesn't have, hehe.Ok, now I'm not bagging the X2 or AMD's top range as they are capable of doing the task required, but the for this test it's important to have Intel as the main test bed since it's got a higher threshold for things tested.
Like I can't imagine games use all the sustained bandwidth yet, but a game playing whilst multitasking should put enough strain to show degradation in available memory data for game textures etc.
If I'm correct from quick calcs, an AMD with top range DDR1 memory running a game at high refresh rates may only have 30meg bandwidth per game frame to play with. On the Intel that is about 55meg per frame. Now these sound like high values but when you consider multitasking and instance demands and gaming hi res textures etc, you start seeing the limitations.
This test fails to test multitasking, it fails to place the strain of FRONTLINE games on the bus, it fails to compare the best of CPU's for the situation.
porkster - Monday, December 5, 2005 - link
If GFX drivers are going to start using multi core cpus and their threads then that will surely increase instance bandwidth on the memory bus, something Intel has trumphs over AMD.I hope the new dual core driver test compares the difference in bandwidth use and what effect that has between AMD and Intel based system.
If you own a AMD without the top notch expensive ram then you maybe trapped is a past era as new games demand more bandiwdth and drivers start using more too.
yacoub - Monday, December 5, 2005 - link
So after looking at the charts it looks like if you use singlecore, stay away. Also no word on if they fixed the issue with FEAR that was mentioned last month as NOT being fixed in 5.11.On a side note: Anyone wonder if this is how they will start to push people from single to dual core? (That is, offering improvements for dc at the expense of sc performance.)
wien - Monday, December 5, 2005 - link
There's no other way of doing it really. Multi-threaded code will always run slower than the equivalent single-threaded code on a single-core CPU. (As long as you count threads waiting for disk-IO and stuff like that out of it that is.) If apps are ever to go the multi-threaded route, single-core performance will suffer...stephenbrooks - Monday, December 5, 2005 - link
There's a thing called an "if" statement :) You can write "if (nprocessors>1) {/* Multithreaded code */} else {/* Single-threaded code */}".Questar - Monday, December 5, 2005 - link
BS. There are many apps that are multithreaded that don't take a perf hit on a single cpu.yacoub - Monday, December 5, 2005 - link
Is this the new Cat driver that fixes the FEAR.EXE bug?