VOGONS


Weird Benchmark results, Socket 7

Topic actions

First post, by Pingaloka

User metadata
Rank Member
Rank
Member

Recently fried 2 MOBOS!!! One of them being my beloved 486...another being an Abit Socket 7 I was going to use in the place of the previous one.

So, looking at other MOBOS I have, I decided to try out the following one:

y80s.jpg

Couldn't gather much information about it apart from finding out that it uses an Intel 82430fx chipset

Decided to do some benchmarking with an Intel Pentium 120mhz CPU. The results I got are a little bit weird. Specially when disabling both internal and external cache. It seems disabling external cache has no effect whatsoever?

A) All settings

4q0k.jpg

B) Internal cache disabled, External enabled

i8q6.jpg

C) Internal cache Enabled, External disabled

4q0k.jpg

D) Both caches disabled

ft3r.jpg

Reply 4 of 27, by vetz

User metadata
Rank l33t
Rank
l33t

Disabling L2 cache does not affect the CPU score in Speedsys. You only see results in other benchmark tests.

If you could do the same benchmarks with the CPU at 90 and 100mhz as in this thread that would be appreciated.

Here are the results from that thread (reposted chart). Notice the same score on a board with fake L2 and on one with L2:

486vsP90_Benchmark_v2.PNG

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 5 of 27, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Where's the timing graphs in speedsys which indicate the L1/L2/RAM speeds? This would shed light on if you have fake cache or not.

Plan your life wisely, you'll be dead before you know it.

Reply 6 of 27, by Pingaloka

User metadata
Rank Member
Rank
Member
vetz wrote:
Disabling L2 cache does not affect the CPU score in Speedsys. You only see results in other benchmark tests. […]
Show full quote

Disabling L2 cache does not affect the CPU score in Speedsys. You only see results in other benchmark tests.

If you could do the same benchmarks with the CPU at 90 and 100mhz as in this thread that would be appreciated.

Here are the results from that thread (reposted chart). Notice the same score on a board with fake L2 and on one with L2:

486vsP90_Benchmark_v2.PNG

Sure Vetz, I will.

Ok here are cachechk results (which I don't know how to read BTW) Is that a Yes or a No?:

xnv4.jpg

Reply 7 of 27, by elianda

User metadata
Rank l33t
Rank
l33t

I don't see anything weird in your results.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 8 of 27, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

Use 3dbench to test if the cache is working!

For slow results use version 1.0 and for fast results use 1.0c.

You can find all the files in my first post here:

3DBENCH CPU Benchmark Database)

PCPBENCH /VGAMODE also works well.

The L2 cache benefits from higher FSB, so you will see the results (with L1 off and L2 on) scale nicely with FSB. If 66 MHz is max you will roughly get:

386DX with L1 off and L2 off
Fast 486DX / slow DX2 with L1 off and L2 on

If you can use a MMX chip you will get a higher performance (fast 386DX and 486DX2)

K6 is the "slowest" chip with caches turned off. Cyrix is just as fast as a Pentium MMX.

Because your board isn't a modern Super Socket 7 board all your scores might be a bit lower, but should be close enough to my list above.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 9 of 27, by vetz

User metadata
Rank l33t
Rank
l33t
Mau1wurf1977 wrote:

Because your board isn't a modern Super Socket 7 board all your scores might be a bit lower, but should be close enough to my list above.

This is exactly why I want to benchmark all socket 5 and socket 7 boards to make a full comparison. I would expect very noticable difference between the MVP3 chipset and 430FX! The differences on the different chipsets are huge as can be seen in the thread I linked to. I don't have a 430FX board yet, so if Pingaloka could run the tests on his board that would be appreciated 😀

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes

Reply 10 of 27, by Pingaloka

User metadata
Rank Member
Rank
Member
vetz wrote:
Mau1wurf1977 wrote:

Because your board isn't a modern Super Socket 7 board all your scores might be a bit lower, but should be close enough to my list above.

This is exactly why I want to benchmark all socket 5 and socket 7 boards to make a full comparison. I would expect very noticable difference between the MVP3 chipset and 430FX! The differences on the different chipsets are huge as can be seen in the thread I linked to. I don't have a 430FX board yet, so if Pingaloka could run the tests on his board that would be appreciated 😀

Please tell me which programs and in what conditions should I run the tests vetz.

Reply 11 of 27, by rgart

User metadata
Rank Oldbie
Rank
Oldbie

It looks like your loading autoexec.bat and config.sys before doing your tests.

I always test without loading anything to interfere - holding shift on boot for example and bypassing system files.

If its a windows PC I hit F8 and go to command prompt.

How did you fry the 2 boards?

Reply 12 of 27, by Pingaloka

User metadata
Rank Member
Rank
Member
rgart wrote:
It looks like your loading autoexec.bat and config.sys before doing your tests. […]
Show full quote

It looks like your loading autoexec.bat and config.sys before doing your tests.

I always test without loading anything to interfere - holding shift on boot for example and bypassing system files.

If its a windows PC I hit F8 and go to command prompt.

How did you fry the 2 boards?

The first one is not really fried, but its unusable, here's a thread about it:

Broken 486 MOBO?

The second one was plugin an hdd led cable to the MOBO while it was on. 😊
Smoke and smell came out....finito....jalas....se acabó

Reply 13 of 27, by Pingaloka

User metadata
Rank Member
Rank
Member

Ok guys, just got some results:

Intel Pentium 120mhz, FSB 60mhz, , 82430fx chipset

**********3DBENCH**********

A) All settings on: 94.2
B) Internal cache disabled, External enabled: 8.1
C) Internal cache Enabled, External disabled: 104.5!! 😲
D) Both caches disabled: 8.3

*********SPEEDSYS*********

A) All settings on: 88.98
B) Internal cache disabled, External enabled: 8.06
C) Internal cache Enabled, External disabled: 88.99
D) Both caches disabled: 8.14

--------------------------------------

Intel Pentium 133mhz, FSB 60mhz, , 82430fx chipset

**********3DBENCH**********

A) All settings on: 104.2
B) Internal cache disabled, External enabled: 9.0
C) Internal cache Enabled, External disabled: 115.6
D) Both caches disabled: 9.3

*********SPEEDSYS*********

A) All settings on: 98.43
B) Internal cache disabled, External enabled: 8.92
C) Internal cache Enabled, External disabled: 98.43
D) Both caches disabled: 9.02

--------------------------------------

Intel Pentium 100mhz, FSB 50mhz, , 82430fx chipset

**********3DBENCH**********

A) All settings on: 78.4
B) Internal cache disabled, External enabled: (92.7)!!
C) Internal cache Enabled, External disabled: 87.0
D) Both caches disabled: (152.2)!!

*********SPEEDSYS*********

A) All settings on: 74.16
B) Internal cache disabled, External enabled: 6.72
C) Internal cache Enabled, External disabled: 74.16
D) Both caches disabled: 6.79

--------------------------------------

Intel Pentium 90mhz, FSB 50mhz, , 82430fx chipset

**********3DBENCH**********

A) All settings on: 82.2
B) Internal cache disabled, External enabled: 7.9
C) Internal cache Enabled, External disabled: 89.9
D) Both caches disabled: 8.3

*********SPEEDSYS*********

A) All settings on: 66.75
B) Internal cache disabled, External enabled: 7.84
C) Internal cache Enabled, External disabled: 66.75
D) Both caches disabled: 8.08

Last edited by Pingaloka on 2013-06-23, 17:07. Edited 5 times in total.

Reply 14 of 27, by elianda

User metadata
Rank l33t
Rank
l33t
Pingaloka wrote:
Ok guys, just got some results: […]
Show full quote

Ok guys, just got some results:

**********3DBENCH**********

Intel Pentium 120mhz, FSB 60mhz, , 82430fx chipset

A) All settings on: 94.2
B) Internal cache disabled, External enabled: 8.1
C) Internal cache Enabled, External disabled: 104.5!! 😲
D) Both caches disabled: 8.3

*********SPEEDSYS*********

A) All settings on: 88.98
B) Internal cache disabled, External enabled: 8.06
C) Internal cache Enabled, External disabled: 88.99
D) Both caches disabled: 8.14

Ok guys, just got some results:
**********3DBENCH**********

Intel Pentium 120mhz, FSB 60mhz, , 82430fx66 chipset

A) All settings on: 106.4
B) Internal cache disabled, External enabled: 25.5
C) Internal cache Enabled, External disabled: 102.0
D) Both caches disabled: 10.8

*********SPEEDSYS*********

A) All settings on: 89.44
B) Internal cache disabled, External enabled: 24.20
C) Internal cache Enabled, External disabled: 89.31
D) Both caches disabled: 10.48

Additional Notes: Board Shuttle HOT-451 Ver 2.5
Settings: EDO in Bank 1 2x32 MB
RAS Precharge 3
R/W Leadoff 7/5
RAS to CAS 2
Read and Write Burst x2222
256 kB PB Cache

Matrox Millenium with 2 MB expansion.

I would like to note also that your FX board uses 256KB Async. SRAM as Cache instead of PB Cache, which explains the results rather well.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 15 of 27, by Pingaloka

User metadata
Rank Member
Rank
Member
elianda wrote:
Ok guys, just got some results: **********3DBENCH********** […]
Show full quote
Pingaloka wrote:
Ok guys, just got some results: […]
Show full quote

Ok guys, just got some results:

**********3DBENCH**********

Intel Pentium 120mhz, FSB 60mhz, , 82430fx chipset

A) All settings on: 94.2
B) Internal cache disabled, External enabled: 8.1
C) Internal cache Enabled, External disabled: 104.5!! 😲
D) Both caches disabled: 8.3

*********SPEEDSYS*********

A) All settings on: 88.98
B) Internal cache disabled, External enabled: 8.06
C) Internal cache Enabled, External disabled: 88.99
D) Both caches disabled: 8.14

Ok guys, just got some results:
**********3DBENCH**********

Intel Pentium 120mhz, FSB 60mhz, , 82430fx66 chipset

A) All settings on: 106.4
B) Internal cache disabled, External enabled: 25.5
C) Internal cache Enabled, External disabled: 102.0
D) Both caches disabled: 10.8

*********SPEEDSYS*********

A) All settings on: 89.44
B) Internal cache disabled, External enabled: 24.20
C) Internal cache Enabled, External disabled: 89.31
D) Both caches disabled: 10.48

Additional Notes: Board Shuttle HOT-451 Ver 2.5
Settings: EDO in Bank 1 2x32 MB
RAS Precharge 3
R/W Leadoff 7/5
RAS to CAS 2
Read and Write Burst x2222
256 kB PB Cache

Matrox Millenium with 2 MB expansion.

I would like to note also that your FX board uses 256KB Async. SRAM as Cache instead of PB Cache, which explains the results rather well.

Elianda could you develop a bit on that please? What is the difference? Are there many boards that work this way?
How can L2 cache off increase performance?

thanx

Reply 16 of 27, by elianda

User metadata
Rank l33t
Rank
l33t

Ok I found a table that shows the timing quite nicely:
http://www.pcguide.com/ref/mbsys/cache/timing … mparison-c.html

So at 60 MHz FSB the Async. Cache requires 9 cycles, whereas the PB cache requires just 6 cycles. Thats just for the overall performance difference.

Actually I had written here something that enabling Cache introduces also latency for Read/misses and that small benchmark loops are prone to cache sizes.
But the more I think about it, the more unlikely it is. Ok both benchmarks use small loops, but it can't explain the scores completely.
While latency could slow it down for reads, it should still be at least a bit faster than with all caches off.

However, if you show is the memory transfer graphs from speedsys and maybe the results of ctcm7r it could shed some more light on this L2 score ratio.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 17 of 27, by elianda

User metadata
Rank l33t
Rank
l33t

To give Vetz some results:
Same setup, but CPU set to P90

Speedsys 4.78:
CPU 67.09
Average L1 177.97 MS/s
Average L2 90.39 MB/s
Average Mem 66.18 MB/s

PCPBench 13.5
3dbench 1.0: 99.99+
3dbench1.0c: 91.4

Doom: 58.3 (2134/1281)
Quake1 320x200 : 25.9 fps
Quake1 640x480 : 9.7 fps

Last edited by elianda on 2013-06-23, 19:38. Edited 1 time in total.

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 18 of 27, by elianda

User metadata
Rank l33t
Rank
l33t

i430fx_bios0.png

i430fx_bios.png

i430fx_infobox.png

i430fx_cachechk.png

Retronn.de - Vintage Hardware Gallery, Drivers, Guides, Videos. Now with file search
Youtube Channel
FTP Server - Driver Archive and more
DVI2PCIe alignment and 2D image quality measurement tool

Reply 19 of 27, by vetz

User metadata
Rank l33t
Rank
l33t
elianda wrote:
To give Vetz some results: Same setup, but CPU set to P90 […]
Show full quote

To give Vetz some results:
Same setup, but CPU set to P90

Speedsys 4.78:
CPU 67.09
Average L1 177.97 MS/s
Average L2 90.39 MB/s
Average Mem 66.18 MB/s

PCPBench 13.5
3dbench 1.0: 99.99+
3dbench1.0c: 91.4

Doom: 58.3 (2134/1281)
Quake1 320x200 : 25.9 fps
Quake1 640x480 : 9.7 fps

Thanks Elandia! What the hell is up with that Doom score? That is much higher than what I got on my VX board and it is higher than what feipoa got in his 686 benchmark (55.3 FPS)

3D Accelerated Games List (Proprietary APIs - No 3DFX/Direct3D)
3D Acceleration Comparison Episodes