VOGONS


Phil's Ultimate VGA Benchmark Database Project

Topic actions

Reply 460 of 495, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++
James-F wrote:

Phil I see a new version of the Benchmark on your site, and also a new 486 video.

Thanks, updated the video thread. Yea the benchmark pack is just an updated version, not published yet, but anyone that stumbles across it can test it 😀 There are new tests, but all the existing ones are identical with the old ones, so results can be taken from both.

YouTube, Facebook, Website

Reply 461 of 495, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++

So for the time being, this is how you can add results to the sheet while I check out the use of forms.

You need to be signed in with a Google account. Then on the top left of the sheet, you can request access and I'll grant it soon enough.

Let's see how this works 😊

EDIT: Also added my first result in a long time! It's a 1 GHz machine with a VIA Eden ULV Esther processor. Quite snappy with a total FPS of 414.81.

Attachments

  • Request Access.JPG
    Filename
    Request Access.JPG
    File size
    45.34 KiB
    Views
    1936 views
    File comment
    Request access to the document
    File license
    Fair use/fair dealing exception

YouTube, Facebook, Website

Reply 462 of 495, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie

Glad the database is back to being editable now Phil.
Also nice to see you stepping into the VIA CPU waters, what are your plans for the little beast?
Best regards,
Chris

286 20MHz,1MB RAM,Trident 8900B 1MB, Conner CFA-170A.SB 1350B
386SX 33MHz,ULSI 387,4MB Ram,OAK OTI077 1MB. Seagate ST1144A, MS WSS audio
Amstrad PC 9486i, DX/2 66, 16 MB RAM, Cirrus SVGA,Win 95,SB 16
Cyrix MII 333,128MB,SiS 6326 H0 rev,ESS 1869,Win ME

Reply 463 of 495, by PhilsComputerLab

User metadata
Rank l33t++
Rank
l33t++
BSA Starfire wrote:
Glad the database is back to being editable now Phil. Also nice to see you stepping into the VIA CPU waters, what are your plan […]
Show full quote

Glad the database is back to being editable now Phil.
Also nice to see you stepping into the VIA CPU waters, what are your plans for the little beast?
Best regards,
Chris

To be honest, I haven't done any research about this platform! Tell me what to do! 😊

The board is special, there will be a video, but if there is anything else interesting I can look into that also. Still waiting for a PCI graphics card to arrive, the onboard GPU didn't work well for me.

YouTube, Facebook, Website

Reply 464 of 495, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
PhilsComputerLab wrote:
BSA Starfire wrote:
Glad the database is back to being editable now Phil. Also nice to see you stepping into the VIA CPU waters, what are your plan […]
Show full quote

Glad the database is back to being editable now Phil.
Also nice to see you stepping into the VIA CPU waters, what are your plans for the little beast?
Best regards,
Chris

To be honest, I haven't done any research about this platform! Tell me what to do! 😊

The board is special, there will be a video, but if there is anything else interesting I can look into that also. Still waiting for a PCI graphics card to arrive, the onboard GPU didn't work well for me.

I'm wondering how well it would work as a highly compact DOS time machine.

It'd be cool to try a Yamaha PCI card with it both in DOS and Windows. Other things as well. SETMUL compatibility and performance range. Graphics compatibility.

All hail the Great Capacitor Brand Finder

Reply 468 of 495, by kithylin

User metadata
Rank l33t
Rank
l33t

I know this is now an old thread, but... how do you convert dooms' realtics to FPS exactly? I used to type in realtics into the google sheet thing but it's not editable. So I dunno.. Wondering what math you use for it anyway?

Reply 471 of 495, by Kamerat

User metadata
Rank Oldbie
Rank
Oldbie

Added some results for one of my rigs at the "Emulators" and "Overclocked" pages. The results for 3DBENCH2 and PCPBench FPS are almost the same between DOSEMU, NTVDM and DOS 7.1.

DOS Sound Blaster compatibility: PCI sound cards vs. PCI chipsets
YouTube channel

Reply 473 of 495, by clueless1

User metadata
Rank l33t
Rank
l33t
infiniteclouds wrote:

I don't understand -- how is real DOS being run on these modern CPUs like i7s? Virtual machines? I can't find anything on this site or youtube with regards to this.

You can transfer a hard drive with DOS installed to any PC and it will probably boot to the DOS prompt just fine. That's how I do it when running DOS benchmarks on various machines old and new. All the benchmarks are already on the the HDD, so just plug it in, boot off of it, and start benchmarking.

The more I learn, the more I realize how much I don't know.
OPL3 FM vs. Roland MT-32 vs. General MIDI DOS Game Comparison
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks

Reply 475 of 495, by Kamerat

User metadata
Rank Oldbie
Rank
Oldbie
infiniteclouds wrote:

Interesting. I don't suppose you could get sound working if your machine had a SB Live! and the drivers were already set up on the DOS install?

No, Ensoniq/Creative cards like the Live! stopped working on ICH6 and up. Yamaha YMF7xx (with the DSDMA TSR) and Aureal Vortex might work on earlier Core i7 systems (don't know if it works on later PCIe to PCI bridged boards, I guess no).

I had my YMF744 running on a first generation i7 system using the TSR.

DOS Sound Blaster compatibility: PCI sound cards vs. PCI chipsets
YouTube channel

Reply 476 of 495, by AlaricD

User metadata
Rank Oldbie
Rank
Oldbie
infiniteclouds wrote:

I don't understand -- how is real DOS being run on these modern CPUs like i7s? Virtual machines? I can't find anything on this site or youtube with regards to this.

Those modern CPUs still support the original 8086 instruction set.

"The Big Bang. The ultimate hero of low frequency. The divine intergalactical bass drum connecting the tribes of our solar system."
Yello
"Solar Driftwood"

Reply 477 of 495, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie

Nice thread, i looked for something like that even did so my benchmarking, but after i found this.
Thanks to all for results, its interesting material to study.

I have few suggestions to online document to make it more user friendly:
- its really annoying to scroll from left to right multiple times to find about on which HW and which settings test ran. Its should be about videocards!, so why videocard info is in some collumns on the end, same as most performance important setting info like if FastVid or MTRR (it can triple performance so its very important).. were used?
I suggest to move all really performance important data to left and after that add results columns and any other additional nitpicking info with used case color, fan speeds, room temperature (im joking but present detail is forcing me to do it.. there are listed all changed bios settings etc..) .. could right side after results, there is plenty of space for someone would bother to scroll there (for this temporary document is not inportant i just hid them). I need to find some data quick, so i made own copy of sheet from user perspective in end with this view - all important info on 1 screen:
Phils sheet compact view
Maybe Phil can add new Sheet with such view, and people can select which view they like more..

- Really dont care about syntetic results (they are not games, i dont like 3D Mark for modern games too) so i hide them.
- i think that for some long additional info, if someone really feels that is needed we can use cells comments its proven solution, better than super long columns.
- i also merged stock and overclocked results, we are not at Intel headquarters to look at overclocking as something bad, it there is info that over clocking was made (i think that @ XXX MHz symbol which is Phil using in his videos spec slides is far enough) is competently fine have these results in the place in my book.
- all cpu speed such be in MHz no GHz to make possible sorting / charts and work which could be done i with integers / floats -> If we mode MHz metric only to header and not use inside in cells, i would also save some typing, again its proven solution for people which are working with sheets or data often.
- convert all VRAM Sizes to MB and again move metric only to header and able for example sort by VRAM size
- add Quake 640x480 + 1024x768 or 1280x1024 collumns - because present numbers for faster machine, are almost unless explained - bellow..

Results
- i added all my conclusion to this thread and there also some my own new results, i broke PCI cards Quake record:) , which i would like to add to Phils sheet:
Old+Modern videocards pure DOS benchmarking- which one is fastest?, need your numbers + analysis, 320x200 to 1600x1200!
- more details in my thread, but i found out that for any not super slow PCI 2MB+ Videocard, Quake 320x200 its really CPU benchmark not Videocard benchmarch, so from Videocard perspective its almost useless.. It scale with CPU speed with same card up to heaven. With Quake 640x480+ testing results not more scale with CPU power and are really capped by Videocard performance.
Other important thing is that if you have great result with 320x200 (because you measuring mainly CPU perforamnce), it doesnt really mean almost anything for bigger resolution, performance there could be really poor.
My Quake results are strange, but i reach them on 3 machines.. 1024x768 is faster that 640x480, on my X58 Xeon performance in higher resolution is poor regardless of used card.

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.

Reply 478 of 495, by elianda

User metadata
Rank l33t
Rank
l33t

Notes:
- add the end of each column a link to the related entry in the original sheet that shows all data
- give overclocked machines at least a mark, like an asterisk. The reason to split it was not that overclocking is evil, it was that comparability of stock systems is better if it is not mixed. Specifically because overclocked systems have often more settings tuned as listed in the simple view. Especially on older computers this can have a dramatical imapct on performance.
- You are right with Quake CPU vs FPU limited on old systems. However introducing 640x480 means stepping from VGA to SVGA. This means all VGA systems can not run it, where it is mostly a FPU bench. Also there is no data in the current sheet for that.
- Doom alone is a bad choice for VGA performance as it uses ModeX and not the standard modes. The result may not say much about regular VGA performance. Also CPU limited on slower systems.
- When leaving out 3DBench[2] and PCPBench consider renaming the sheet from 'VGA benchmark' to 'Doom/Quake benchmark'.
- It would be also a good opportunity to remove senseless entries, like using FASTVID/MTRRLFBE as performance enhancers on CPUs that don't support it. Like the Pentiums/Mobile Pentiums and/or systems with ISA cards...

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 479 of 495, by ruthan

User metadata
Rank Oldbie
Rank
Oldbie
elianda wrote:

Notes:
- add the end of each column a link to the related entry in the original sheet that shows all data

I will just theorize about it, because my sheet is not synchronized with original, you i make no sense to improve it more, unless Phil will decide integrate such view in original sheet. Maybe better solution would be not hide anything, only just move not crucial columns to right after most import columns, which has to be in left before results itself. Link - im not sure if possible use something like links within multiple sheet to point some exact cell by some formula. If someone know to do it in Google sheet, just tell me, i can be handy.

elianda wrote:

- give overclocked machines at least a mark, like an asterisk. The reason to split it was not that overclocking is evil, it was that comparability of stock systems is better if it is not mixed. Specifically because overclocked systems have often more settings tuned as listed in the simple view. Especially on older computers this can have a dramatical imapct on performance.

Sorry i dont really understand this argument, you telling is that is nothing bad with overclocked results, but you not act like that, i would call it results hypocrisy:) I dont thing that someone is using liquid nitrogen for DOS overclocking. In my view you can identify overclock results by word integrated, in FPU column, which i used instead of Integrated FPU, it was not intentional, but is way how to identify them.

elianda wrote:

- You are right with Quake CPU vs FPU limited on old systems. However introducing 640x480 means stepping from VGA to SVGA. This means all VGA systems can not run it, where it is mostly a FPU bench. Also there is no data in the current sheet for that.

I dont see anything wrong with optimal columns, lots of columns are already optimal from inserted data, its better to have 1 more optimal column, that have only Quake 320x200 column, where result its not really about video cards for 3/4 of inserted results - optimal columns can you give best of both world. I personally not really care about sum FPS result, new column not need to included, its only small adjustment to change formula how you like.. - for example i would really use only Doom and Quake 640 and 1024 for my personal analysis an which videocard to buy decision.

elianda wrote:

- Doom alone is a bad choice for VGA performance as it uses ModeX and not the standard modes. The result may not say much about regular VGA performance. Also CPU limited on slower systems.

Its game which defined whole era and forced people to buy faster 486. But tell me other good DOS game with inbuild or 3rd benchmarking option. 3DBENCH2 FPS has lots of Too fast results its not good too.

elianda wrote:

- When leaving out 3DBench[2] and PCPBench consider renaming the sheet from 'VGA benchmark' to 'Doom/Quake benchmark'.
- It would be also a good opportunity to remove senseless entries, like using FASTVID/MTRRLFBE as performance enhancers on CPUs that don't support it. Like the Pentiums/Mobile Pentiums and/or systems with ISA cards...

Whole 4 benchmarks is still hardly to be fully representative for DOS. In Phils sheet are 4 results from 4 different programs and in my analysis are also 4 results (only 3 from 1 game - Quake). I think that is good to know even that FASTVID/MTRR and others were tested is are not supported on some systems, enhancers are best way how to use HW. In my personal sheet i have results for both with+without enhancers:
https://docs.zoho.com/sheet/published.do?rid= … c62b9&mode=html // Now there is some shortage of Zoho, problem with DNS, if its not working try it later, or use workaround google dns - 8.8.8.8
https://downdetector.com/status/zoho

Im old goal oriented goatman, i care about facts and freedom, not about egos+prejudices. Hoarding=sickness. If you want respect, gain it by your behavior. I hate stupid SW limits, SW=virtual world, everything should be possible if you have enough raw HW.