RAM question for Giga-byte GA-5AX rev 2. (Update 7-9-06)

Discussion relating to Socket 7 hardware.
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

Post by Stedman5040 »

@Jim

Yes Jim I did use the Cache @66 tweak and it does give a big boost with the AX59Pro board. However along the tweak trail with your DFI board as reference my board missed out on some of the gains that you got. I don't know why. Perhaps the bios on your DFI board sets up differently in wpcredit and I am missing some of the original settings. What makes it even worse is that in your original thread you used a K6-III+ @ 600MHz and mine was set at 550MHz.

So I need to find out the source of the difference. Maybe it is all down to memory size. I'll have to look for some compatible 256mb memory.

Stedman
Jim
K6'er Elite
Posts: 1745
Joined: Wed Jan 21, 2004 7:10 pm
Location: Toronto

Post by Jim »

Get Hynix or Crucial or something like that if you can. They support some settings that others won't.
NOTE : The best SDRAM that I have is the 2 sticks in Superpuppy-3. They are Hynix "Low Power" RAM. Although they are PC133 CL3 PC100 CL2 sticks they support some settings that my Hynix PC133 CL2 sticks will not. I think this is because they are low power; but in WPCredit, I find that the power settings are all set by default to the higher level "Normal Power" settings. That is probably why they are able to continue to function under stress that causes other RAM to crap out. Hynix made 256 Meg PC133 CL2 Low Power sticks too. Wish I could find some.

EDIT : IMPORTANT : @ Stedman : If you go to a "guest" post of mine, in the Hardware/Tech Corner section, - Page 2 - Topic - "DFI K6BV3+/66 vs ASUS P5A-B", - (which hardly anybody ever looks at), you will find that KYLE BRANT, the author of K6Speed, posted his WPCredit settings for an FIC board, which also has an MVP3 chipset. You may find that his tweaks work better than mine on your board which is different from both his and mine.

RE-EDIT : Would also appreciate it, Stedman, if you could run the same tests w/ the same settings, (insofar as applicable), on that MSI board; but w/ 256 Meg & 384 Meg of RAM instead of 768. Then edit your post to give the comparison between 256, 384, & 768 Meg of RAM installed. Thanks.
Superpuppy 3
K6-3+ 450 ACZ (6x100)
DFI K6BV3+/66 Rev B2 (2 Meg) w/ 2x28mm Chipset Fans
2x256 Meg PC 133 Hynix SDRAM
1x 20G Maxtor (7200)
2x 80G Maxtor (7200) Ducted w/ 2x486 Fans Mount
52/24/52/16 LG CDR/RW/DVD
8/4/3/12/24/16/32 LG Super Multi
ATI 9000 aiw Radeon AGP
SB Audigy 1 MP3 Sound
CMD 649 IDE Controller
NEC USB 2 Card
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

More MS5169 results

Post by Stedman5040 »

Following the Everest and Sandra results from yesterday I have taken the same set up and benched different amounts of RAM in the MS5169 board. The following results make very interesting reading.

The method used for each run is as follows

1. Fire up Windows
2. Open Wpcredit if applicable and make tweaks
3. Close down Wpcredit
4. Open up Everest and run memory benchmarks 3x for each test
5. Close down Everest
6. Open up Sandra 2004
7. Run Memory bandwidth test until readings are stable. Usually about 5 times
8. Close down Sandra
9. Open up Everest and re run the memory benches x3
10. Close down Everest
11. Shut down computer ready to start again

Set up 1 as follows

K6-III+ @ 550 (5.5x100)
MS5169 Ver4.0 (Ali chipset) 512k L3 Cache
1x 256mb SDRAM High density PC133
Intel i740 graphics
Creative SB128
24x CDROM

Run1.

Default settings as booted up no tweaks other than usual CTU start up. Memory timings at 2-2-4-5

1. Everest MR/MW/ML 331/153/200
2. Sandra start of run 185/177
3. sandra end of run 178/176
4. Everest MR/MW/ML 328/145/202

Run2.

Memory timings changed with wpcredit to 2-2-2-5

1. Everest MR/MW/ML 333/161/199
2. Sandra start of run 183/179
3. Sandra end of run 180/179
4. Everest MR/MW/ML 329/142/198

Run3.

As above but with mixed dram set to [1] with wpcredit

1. Everest MR/MW/ML 331/161/199
2. Sandra Start of run 189/183
3. Sandra end of run 182/180
4. Everest MR/MW/ML 329/141/198

Run4.

As above but also with Fast Dram read enabled with wpcredit

1. Everest MR/MW/ML 351/168/188
2. Sandra start of run 190/185
3. Sandra end of run 188/186
4. Everest MR/MW/ML 345/150/189

Please note that with all the above tests the Sandra readings go down with successive tests and end up with poorer Everest readings the second time around. Just look at that drop in Memory write (more than 10%)


Set up 2 as above but with

3x 256mb SDRAM high density PC133


Run1.

As run 1 above.

1.Everest MR/MW/ML 332/155/200
2.Sandra Start of run 192/183
3. Sandra end of run 204/200
4.Everest MR/MW/ML 332/176/204

Run2.

As run4 above ie all tweaks

1. Everest MR/MW/ML 352/170/188
2. sandra start of run 195/191
3. Sandra end of run 212/208
4. Everest MR/MW/ML 352/182/192

Clearly more memory changes the trends in this testing. In this case sandra bandwidth results go up and the subsequent Everest results also go up

So as Jim has noted on many occasions before lots of ram improves Everest MW results after repeated testing with Sandra memory bandwidth benchmarks. Has anyone seen the opposite effect as seen above with lower amounts of memory??

Comments please??

Stedman
Jim
K6'er Elite
Posts: 1745
Joined: Wed Jan 21, 2004 7:10 pm
Location: Toronto

RE: More MS5169 results

Post by Jim »

Thanks for the confirmation Stedman. The bit about the falling results w/ small ram installation is new to me personally; though I seem to recall that others here have reported it; anyway - I got sold on more RAM a long time ago..

F.A.Q. : "What is the best thing I can do to improve the performance of my machine?"
ANSWER : "Add more RAM".

F.A.Q. : "What is the second best thing I can do to improve the performance of my machine?"
ANSWER : "Add more RAM".

F.A.Q. : "What is the third best thing I can do to improve the performance of my machine?"
ANSWER : "Add still more RAM".

BTW , any luck w/ CPU Cool? One other thing, I hate to ask; but Peter is going to want to know : "What happens w/ 768 installed if you disable your mobo cache and use his other bios setting then reboot before running the same tests?"

EDIT : Given those falling results w/ small RAM installations, I'd like to know myself. Could be Peter is right, and results will continue to rise w/ cache disabled, when 768 is installed, though I doubt it.

EDIT : BTW, I will post a chart of my MVP3 WPCredit settings, (including defaults), for you as soon as I get that machine back together after its rebuild.
Superpuppy 3
K6-3+ 450 ACZ (6x100)
DFI K6BV3+/66 Rev B2 (2 Meg) w/ 2x28mm Chipset Fans
2x256 Meg PC 133 Hynix SDRAM
1x 20G Maxtor (7200)
2x 80G Maxtor (7200) Ducted w/ 2x486 Fans Mount
52/24/52/16 LG CDR/RW/DVD
8/4/3/12/24/16/32 LG Super Multi
ATI 9000 aiw Radeon AGP
SB Audigy 1 MP3 Sound
CMD 649 IDE Controller
NEC USB 2 Card
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

Kyle Brant settings

Post by Stedman5040 »

@Jim

OK I hooked the AX59PRO back up with the usual set up

K6-III+/450 @ 550 (5.5x100)
3x128mb Hynix CL2 pc133
Geforce2Ti
Creative SB Live 5.1
WD 5,400rpm 20G HDD

I ran Everest with all of Jim's tweaks for the DFI board and got the following results

MR/MW/ML (265/142/250)

I then applied the few applicable changes as shown in the Kyle Brant dump of wpcredit for the Via 503+. The results I got with Everest were slightly better with the memory latency, but no big change.

MR/MW/ML (266/142/243)

I really need to try this set up with more memory to see if I get a boost in performance. This board is not getting anywhere near Jim's figures for his DFI board.

Stedman
User avatar
KGB
K6'er
Posts: 98
Joined: Mon Apr 07, 2003 7:55 pm

Post by KGB »

Hey guys,
Sorry for the non-update, having tremendous toruble with the graphics card not functioning after installing its drivers you can look at this url to see a topic on what sort of problem im having, its so strenuous to figure out a solution http://forum.matrox.com/mga/viewtopic.php?t=4330

Haven't had a chance to test anything. Im not promising anything, but it just infuriates me with the trouble im encountering with a product of a company i used to admire :(
-K6-2 550 Gigagyte GA-5AX (5.2) 1x512MB GeForce 2 Ti500 64MB
-Pentium III 933 Asus P3C-L 2x128MB GeForce 4 Ti4200 128MB DDR
-Pentium III 750 Asus P3B-F ATi Rage128 GL 32MB
-Celeron 1.2 Asus TUSI-M 2x128MB
-Pentium IV 1.5 DELL 2350 1x128MB
-Athlon 900 Asus K7M 1x256MB 2x128MB GeForce 3 64MB
-AthlonXP-M 2400+ Asus K7V880 2x512MB HIS Radeon HD3850 512MB DDR
Jim
K6'er Elite
Posts: 1745
Joined: Wed Jan 21, 2004 7:10 pm
Location: Toronto

Post by Jim »

I am still down, using Uptown's machine for this, don't like Matrox myself, too much trouble with them, - shutdown problems, - driver problems, etc. I do have a fairly comprehensive set of drivers for various Matrox cards though; and when I am back up can e-mail what I have to you if you want them; but if was me I'd switch to ATI.
Superpuppy 3
K6-3+ 450 ACZ (6x100)
DFI K6BV3+/66 Rev B2 (2 Meg) w/ 2x28mm Chipset Fans
2x256 Meg PC 133 Hynix SDRAM
1x 20G Maxtor (7200)
2x 80G Maxtor (7200) Ducted w/ 2x486 Fans Mount
52/24/52/16 LG CDR/RW/DVD
8/4/3/12/24/16/32 LG Super Multi
ATI 9000 aiw Radeon AGP
SB Audigy 1 MP3 Sound
CMD 649 IDE Controller
NEC USB 2 Card
DonPedro
K6'er Elite
Posts: 578
Joined: Wed Jul 27, 2005 2:11 pm

Post by DonPedro »

@stedman

thanks for your numbers.

question #1: did you testing with cache en- or disabled?
question #2: I wonder how did you manage to start with an everest latency of 200 without any tweaks. that is an amazing good score!!
question #3: could you take a screenshot of your wpcredits window (no manual tweaks applied) and post it here?

@jim

I agree with you that the system performs better with more ram installed - BUT only as long as windows has to swap memory to and from the disk because "physical" memory is scarce. once all programs running fit into "physical memory adding more ram does NOT improve performance. what it does is pleasing some specific artificial benchmark routines. your car won't run faster because you add a fifth tyre.

that leads me to the question: why do some benchmark programs show better results if (allegedly) they fit completely into physical ram (together with the ram they need for their data to perform their bench-tasks on) given that you put additonal (abundant) ram into the system and the bench-runs are carried out with onboard-cache DISabled? If they then show better results then what are they actually measuring? where does the speed gain come from? I think that THAT behaviour is strange and shows that the benchmark is somehow faulty by construction.
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

Back to the P5A

Post by Stedman5040 »

Here are some figures using the Ver1.04 P5A board. The set up is as follows

K6-2+/550 @ 617 (5.5x112)
512Mb pc133 cl3 Hynix ram (2x256mb)
Asus P5A (ver1.04) (cache on)
WD204 20GB 5400rpm HDD
Voodoo3500 AGP graphics
Hercules muse lt sound card
Windows 98se

Everest results from straight boot up with no memory tweaks

MR 380
MW 190
ML 173

2004 Sandra memory bandwidth

ALU 217 rising to 233
FPU 214 rising to 229

Beats the Aladdin7 reported results on Sandra by a full 10% or more.

Everest results after Sandra shows no gains

MR 380
MW 192
ML 170

Not bad results and pretty cool to see the K6-2+/550 at 617MHz 8) . This seems a bit odd to me however as Asus report in the MB fsb settings frequencies of 105, 110, 115 etc and not 112. I'm getting the 112 MHz from the 110MHz setting. It is also nice to see that the onboard cache is running at 112MHz.

I may try some more aggressive settings from within Wpcredit, but I am not sure that the ram will be up to it.

Stedman.
DonPedro
K6'er Elite
Posts: 578
Joined: Wed Jul 27, 2005 2:11 pm

Post by DonPedro »

:D and :o and "there is no icon for jaws dropped" !

I am really surprised that the ali-5 can handle 112mhz fsb. what is left now is how about stability? everest does not put a lot of stress on the system. could you please try the tool "cpu hot tester" (lite) for that purpose.

It does a more sound benchmark (cpu raw calculation power, memory speed, and combinations of it - sorting routines and other stuff) and is in my opinion a very good means to see effects of tweaking on system performance (because it has a more "real world" approach).

it takes about 4-5 minutes to run the benchmark. you can download the tool right here.

I doubt that you can successfully run the osmark-bench .... but let's see first what numbers you get on hot-cpu-tester!!

btw, from what you quote your system is not equipped with a lan-card, right?
Jim
K6'er Elite
Posts: 1745
Joined: Wed Jan 21, 2004 7:10 pm
Location: Toronto

Post by Jim »

@ Stedman: Those are very similar numbers to what I get from Superpuppy-2 when it is working. I suspect that if you up the RAM to 768 Meg, you will get the rising Everest result after running Sandra. Beware of OS Mark. It takes upwards of an hour to run; and it keeps the CPU at the redline for the whole time. If there is any instability present whatsoever, you will wind up w/ massive file corruption. I think the P5A-B cache is good for 112, though I don't know about 115. Will eventually find out once I have SP-3 finished and back up; then I can get serious about SP-2.
Superpuppy 3
K6-3+ 450 ACZ (6x100)
DFI K6BV3+/66 Rev B2 (2 Meg) w/ 2x28mm Chipset Fans
2x256 Meg PC 133 Hynix SDRAM
1x 20G Maxtor (7200)
2x 80G Maxtor (7200) Ducted w/ 2x486 Fans Mount
52/24/52/16 LG CDR/RW/DVD
8/4/3/12/24/16/32 LG Super Multi
ATI 9000 aiw Radeon AGP
SB Audigy 1 MP3 Sound
CMD 649 IDE Controller
NEC USB 2 Card
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

Post by Stedman5040 »

@DonPedro

OK, Yes I do have a LAN card it is basically a RTL8139.

I have run the hotcputester through with no problems and the scores are as follows.

Total score 1156

Metamark to MHz ratio 1.87
----------------------------------------
integer instructions 197 M instructions/s
Floating point instuctions 77 M instuctions/s
MMX instuctions 416 M instructions/s
SSE N/A
SSE2 N/A
3D Now! instructions 379 M instuctions/s
----------------------------------------
Memory fillrate memcpy method 223 MB/s
Memory fillrate C++ method 219 MB/s
Memory rep movsd 220 MB/s
Memory FPU 64 bit 218 MB/s
Memory assembly 32 bit 215 MB/s
Memory fillrate assembly 64 bit 222 MB/s
Memory fillrate MMX 32 bit 226 MB/s
Memory fillrate MMX 128 bit 223 MB/s
-----------------------------------------
Insertion sort 8K Items/s
Shell sort 290 K Items/s
Merge sort 536 K Items/s
Heap sort 298 K items/s
Quick sort 1145 K Items/s
Large objects sort (quick sort method) 246 K Items/s
-----------------------------------------
Fast Fourier Transform (Standard) 497 KB/s
Fast Fourier Transform (Optimised) 512 KB/s
-----------------------------------------

How do these results stack up against your results?

Stedman :?:
DonPedro
K6'er Elite
Posts: 578
Joined: Wed Jul 27, 2005 2:11 pm

Post by DonPedro »

voila some different setups (fsb speed, cpu mhz) I have collected benches on in the past.

don't mind the different colours. they have a meaning not applicable in this specific comparison.

what is remarkable here is that my k6-III@550 numbers namely the fast fourier transformations are way above yours. I don't have an explanation for this.

as far as I can remember I always had the hw-multiplier set @ 5.5 and changed it via k6speed when needed.
Attachments
comparison k6-3+ vs k6-2+ ali5 ga5ax 256mb cache off nt4 gf4mx460.png
comparison k6-3+ vs k6-2+ ali5 ga5ax 256mb cache off nt4 gf4mx460.png (25.6 KiB) Viewed 12336 times
User avatar
Stedman5040
Veteran K6'er
Posts: 271
Joined: Mon Sep 05, 2005 4:22 pm

Post by Stedman5040 »

@Don Pedro

Thanks for the figures. I guess some of the differences are down to the Graphics card. I am now running OSMark on the Asus P5A @ 112MHz fsb. About half way through the second iteration and so far no problems.

Stedman
Jim
K6'er Elite
Posts: 1745
Joined: Wed Jan 21, 2004 7:10 pm
Location: Toronto

Post by Jim »

@ Stedman: Well how did it go w/ OS Mark? - Proved to be a lot of trouble for me.

@ Peter: Don't want to get acrimonious here; but it seems to me that OS Mark is :
1) Designed more to find any weakness in the system than it is to report its performance. Using it on a system that is highly tweaked; will in my experience, trash the OS installation. That in itself is a useful function, because it enables the user to determine the highest level of performance at which the user can expect absolute reliability from the machine. The downside to it is that he, or she will have to reinstall his , or her operating system several times to arrive at what the acceptable tweaks are. As such it is best used on a fresh install with minimal software installed so as to limit the amount of work involved in doing OS reinstallations after trying tweaks that are too aggressive.

2) Not a particularly good tool for determining performance. I say this because the "official run" performs a number of tests 3 times, then takes the best result for each test to calculate the "official score". To determine the veracity of these results, I ran a number of tests, (which you will find in the OSMark results I posted here), 20 and 30 times in immediate succession. In the maze threads test for example, the scores varied from 501 to 551, with the bulk of the results being in the 530s range. The pattern of the results however was completely random; and choosing 3 of them at random as "official run" results will result in varying "official scores" for the particular test. One person may luck out and get the 1 in 30 chance that the 551 score will come up; but others will not. When I get Superpuppy 3 back up, I will run the same test again 100 times to see if that results in an even wider range of result scores. Needless to say the repeatability of scores generated by this type of testing is not going to be very high for the individual tests, although because of the large numbers of different individual tests performed the overall "official score" may average out approximately the same.
Superpuppy 3
K6-3+ 450 ACZ (6x100)
DFI K6BV3+/66 Rev B2 (2 Meg) w/ 2x28mm Chipset Fans
2x256 Meg PC 133 Hynix SDRAM
1x 20G Maxtor (7200)
2x 80G Maxtor (7200) Ducted w/ 2x486 Fans Mount
52/24/52/16 LG CDR/RW/DVD
8/4/3/12/24/16/32 LG Super Multi
ATI 9000 aiw Radeon AGP
SB Audigy 1 MP3 Sound
CMD 649 IDE Controller
NEC USB 2 Card
Post Reply