VOGONS


The Ultimate 486 Benchmark Comparison

Topic actions

Reply 100 of 280, by elianda

User metadata
Rank l33t
Rank
l33t

One thing I would add for my VLB system is, that the minimum reachable L2-Cache timing strongly depends which VLB cards you have plugged.
So f.e. the VLB ET4000/W32i is faster than a SPEA VEGA if you bench just the card, but with the VEGA I can run tighter L2-Cache timings that overall speed up the systems more than using the ET4000 in games.
This is with the VEGA:

and this with the ET4000:

also note that the Speedsys reports for the VEGA 8.8 MB/s and 22.8 MB/s for the ET4000 the difference in actual games can not be reproduced.
The VEGA reaches it's maximum performance already with 16 bit accesses compared to the ET4000 with 32 bit accesses according to vidspeed4.

Last edited by elianda on 2013-12-30, 04:22. 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 101 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I'd have never thought there would be a cache timing dependence on graphic card selection. This opens a whole new avenue for exploration. Perhaps it is a VLB-only occurrence?

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

Reply 102 of 280, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

That's an interesting discovery, but I've never had a problem running my cache at the fastest timings with ET4000W32P in place. I'm using a Hercules Dynamite Power. The card also works very well at 50MHz. Maybe it depends on the quality of the motherboard and graphics card too, not just the chipsets.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 104 of 280, by Mau1wurf1977

User metadata
Rank l33t++
Rank
l33t++

We could need some "exotic" benchmark results for my project: Phil's Ultimate VGA Benchmark Database Project

You can use whatever software utilities you want (for setting registers, speeding up graphics), just document it in the relevant sections 😀

Cyrix, Pentium Overdrive and Pentium Pro would be very good additions.

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

Reply 106 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Graphs contained in the Ultimate 486 Benchmark Comparison have made it to the reference material for an eBay auction. I don't really think it helps the sale of this Evergreen upgrade though.
http://www.ebay.com/itm/181365938124

Attachments

  • Filename
    U4BC_on_eBay.png
    File size
    205.23 KiB
    Downloads
    No downloads
    File license
    Fair use/fair dealing exception

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

Reply 108 of 280, by sliderider

User metadata
Rank l33t++
Rank
l33t++
feipoa wrote:

Graphs contained in the Ultimate 486 Benchmark Comparison have made it to the reference material for an eBay auction. I don't really think it helps the sale of this Evergreen upgrade though.
http://www.ebay.com/itm/181365938124

Even with a socket interposer attached, I don't think it's worth $60. For $60 you can get a socket 3 motherboard with an AMD 5x86 chip and RAM already installed.

Reply 109 of 280, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t
sliderider wrote:

Does anyone know if the 5x86 made by SGS-Thomson supports a 1x multipler? Which multipliers are supported?

All Cyrix based 5x86 chips can do 1X as far as I know, it it needs to be done in software. I can switch my IBM 5x86C into 1X mode with CTCHIP.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 110 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

This post is a note to self

Next revision should include:

1) Run all chips with Biostar MB8433-UUD (UMC 8881/8886) to avoid the auto 2/3 PCI downclock implemented by the M919 when FSB = 40 MHz.

2) Set L2 cache to 256KB double-banked to allow for maximum cache timings. Use single stick of 32 MB memory to ensure all RAM cached.

3) Use Windows95 OSR2 instead of Win98SE.

4) Add GLQuake and Quake II GLmode. Retest with 64 MB RAM and L2 in WT mode to ensure 32 MB RAM is sufficient.

5) Add memory write test from Cachechk.

6) 486 CPUs to add: UMC U5SX-33, UMC U5SX-40, AMD X5-150, Intel DX2-133, AMD DX2-133, Intel 486 DX-50, Intel 486 DX4-100-WT, AMD DX4-100-8KB-WB, AMD DX2-66-16KB-WB, Intel DX2-66-8KB-WB, Intel 486SX, TI 486SXL2-66, P75/90/100

7) 386 CPUs to add: Intel 386-33, Intel 386-40, Cyrix DLC-33, Cyrix DLC-40, Cyrix DRx2-66, TI SXL-40, TI SXL2-50, TI SXL2-66 (unlikely overclock), 386-to-486 interposer with some select socket 3 CPUs,

7b) test up to P54C-100. Add NexGen if possible. Maybe test up to P54C-133 and include K5, MediaGX, and others.

8) Possibly switch video card to ATI Rage 128 VR for increased OpenGL support and speeds. Note the Rage 128 VR Win9x drivers work with all CPUs. For WinNT4, 128 VR drivers work for non-Cyrix-5x86 chips only.

8b) Possibly switch to Matrox G200 w/Voodoo2 for glide benchmarks

8c) Possibly switch to S3 Virge GX2 w/Voodoo2 if I do not care about D3D results.

9) Rerun all tests with DTK PKM-0033S (SiS 496/497) and Voodoo3, Banshee, or ATI Rage 128 VR. Or Select a single CPU, e.g. DX4-100-WB and run the SiS 496 tests. Similarly, run tests with a SiS 471 VLB board w/DX4-100-WB.

10) Eliminate uncommon benchmark suites and add more game-related benchmarks, if they exist. See 11)

11) Keep eye out for other suggestions for improvement, esp. concerning which benchmarks/games to test. - Possibly add game benchmarks "Incoming" and "Forsaken". Also could add Duke3D and Outlaws and check the frame rate counter under identical conditions. Include D3D and Glide results to software results when possible. Include PCPBench VGA and SVGA modes, Topbench, Vspeed, mvspeed, Wing Commander III, Chris3D. Will Tom2D run? Does Unreal Tournament, MDK1, Magic Carpet, GTA, Descent I/II, etc have a fps counter?

12) Run DOS benchmarks with Branch Prediction enabled as this is a stable DOS setting for Cyrix/IBM 5x86 chips.

I will not be doing this anytime soon.

Last edited by feipoa on 2017-03-15, 12:45. Edited 9 times in total.

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

Reply 111 of 280, by vetz

User metadata
Rank l33t
Rank
l33t

Maybe add that Doom & Quake benches to be done with Phil's Ultimate VGA Benchmark settings to get comparable results in those games?

Why do you want to switch video card?

Remember, I've also added all the Winstone CD versions I could find cheaply on Ebay to Vogonsdrivers (Winstone 96 is available from Archive.org):
http://www.vogonsdrivers.com/index.php?catid=58
If you run WInstone 97 that bench works in Win98 as well 😀

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

Reply 112 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Downloaded all those Ziff-Davis benchmarks as well as all those mentioned in that benchmark archive thread.

I was thinking of using Phil's settings for those particular benchmraks. I need to look at which settings he has set first.

I was thinking to switch the graphics card to the ATI Rage 128 VR so that I can run Quake II in GL Mode within Windows 9X instead of needing to boot into Windows NT 4.0. I was unsuccesful in getting the Matrox G200 drivers to work with Quake II GL Mode in Win9x; only the NT 4.0 drivers seem to work with Quake II GL Mode. My main interest is GLQuake and I have assumed that if Quake II GL doesn't work with the Matrox G200 drivers, then GLQuake also will not work. I still need to verify this though. Refer to the Modern Graphics on a 486 thread, Modern graphics on a 486

Users here seem to give the most weight to game benchmarks rather than synthetic ones. Are there any other game benchmarks I can run? I was thinking about MDK2, but that might be a little too modern. Duke Nukem 3D - does that have a timedemo?

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

Reply 116 of 280, by computergeek92

User metadata
Rank Oldbie
Rank
Oldbie

Say, how would a Pentium 75MHz and a Pentium 90MHz rate on integer and fpu scores vs a 486? I own both of these chips.

Dedicated Windows 95 Aficionado for good reasons:
http://toastytech.com/evil/setup.html

Reply 117 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++
computergeek92 wrote:

Say, how would a Pentium 75MHz and a Pentium 90MHz rate on integer and fpu scores vs a 486? I own both of these chips.

Please refer to the Ultimate 686 Benchmark Comparison. I beleive I have some high-end 486's on that comparison as well.

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

Reply 118 of 280, by Mithloraite

User metadata
Rank Member
Rank
Member
feipoa wrote:
This post is a note to self […]
Show full quote

This post is a note to self

Next revision should include:
...
3) Use Windows95 OSR2 instead of Win98SE.

May I ask why? Is it any faster? (W95 OSR2 vs. 98SE)

Perfection is the key. Fatality is the key. (c)

Reply 119 of 280, by feipoa

User metadata
Rank l33t++
Rank
l33t++

I found game and synthetic benchmarks to yield equal scores in Win95c and Win98SE. The desire to switch to Win95c is because this is the OS I have installed on my 486 systems now and being more period correct adds a little to the nostalgia factor.

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