Does anybody have stable working full speed v56k card?

  • Hello. Anybody have this card that works good at 66mhz bus? Think, many of you know that vsa100 chips hardswitched to 33mhz bus on reworked v56k cards, so exising benchmarks exposing low speed. Overclocked v5 5500 agp will outperform such v56k (at least in heavy textured games or benchmarks). So, just wondering how fast could be v56k card at it prime? Is there any solid source info about this?


    may be someone could do simple benchmark on reworked and non reworked card?

  • Thats a interessting question! So i asked myself too. For example on YouTube is a 56K showing Doom3. It runs real slow.

    Using my regular Voodoo5 with little oc (194Mhz) it doubles the Framerate in Doom3.

    I also tried a Voodoo4 and the single VSA100 runs like a V56K. There are many limitations, that are not the fault of 3d Chip.


    Black Mesa 2012 runs on V4 like V5 5500, maybe 10% slower in some cases.

    On heavy demandig Games, i See no real advantage in SLI. No using of higher Fillrate for more Speed.

  • yep, i mean difference between reworked and not reworked cards. and what exactly do 33/66mhz switch in vsa chip, how it works. pci-pci bridge operates at fixed 1:1 ratio and can't provide 33mhz secondary bus clock since being clocked from 66mhz agp clock. so i'd like to figure out what exactly do this rework and how it affects performance of v56k

  • Anthony , now I catch you- you mean if there is difference performance between to set up 6000 66MHz instead of 33MHz (disabling Hank's rework)- answer is NO. Years ago I did many test using 6000s with Hanks Rework and without it. Hanks rework sets up VSA100 as PCI 33MHz instead 66MHz but talking in a global performance there is not impact.


    I did similar test doing Hanks rework on V5 AGP and sets up 33MHz instead of 66MHz, performance was the same even disabling AGP capabilities and seated up V5 AGP as PCI with same performance.


    I used 3Dmark2001 since the difference in performance between 33 and 66MHz is very noticeable using T&L by Software- always same performance.


    Across of an AGP2PCI adapter you can check the difference performance between 33 and 66MHz in any VSA100 card. Difference is very important when emulation via SW is present.


    Regards,

    Oscar.

  • I think that cannot be entirely correct. We had some benchmarks here that showed huge advantages for AGP V4 outperforming V5 being bottlenecked by 'PCI' bus. Will check if i find those...will be in german though.

  • I think that cannot be entirely correct. We had some benchmarks here that showed huge advantages for AGP V4 outperforming V5 being bottlenecked by 'PCI' bus. Will check if i find those...will be in german though.

    Ich habe die selben Beobachtungen gemacht.

  • thanks for that info. anyway, i still curious about what exactly do 33/66mhz switch, no good explanation in available documentation

  • I think that cannot be entirely correct. We had some benchmarks here that showed huge advantages for AGP V4 outperforming V5 being bottlenecked by 'PCI' bus. Will check if i find those...will be in german though.

    Yes, correct. Using a V4 AGP 66MHz and V5-5500 PCI 33MHz in some special game and test V4 will perform better.


    Just Anthony wanted to know if the switch 33/66MHz on VSA100 across strapping really affects on performance as it affects using using a pci bus 33MHz VS AGP 66MHz.


    Seems such switch should do same difference between pci 33 vs agp 66MHz but in reality there is not difference.

  • i didn't get the idea of this 33mhz strap. regarding elimination of those serial termination resistors, it may make a sence, without resistors clock signal will be more sharp, so if these track routed without solid reference it may help. but what staying behind this 33mhz strap?

  • i will answer for myself. 33 mhz strap does greater setup time. in theory, it may help if you want to put voodoo agp card in bx system for instance, and push it beyond limits, for 89mhz agp clock or more. thats my thoughts

  • Maybe it's still interesting? osckhar

    I will check it- do you know any special test for checking it?

    You can measure the bus bandwidth with Final Reality benchmark.


    My result with V5-5500 AGP, Athlon XP @2GHz, AMD761 Chipset, AGP 2x @66MHz:


    2843-bus-transfer-rate-66-png


    Owner of YT-channel 3dfxhistory, with V5-6000 AGP, Athlon XP @2GHz, AMD761 Chipset, "probably" AGP2x @66Mhz:



    YT-Link: https://www.youtube.com/watch?…L971E523111DF5FCE&index=1


    Theoretically 266Mb/s is possible at AGP 2x / 66 MHz.

    Die Pixel beschleunigt, die Kanten gefixt, lang leben die Karten von 3dfx! :spitze:

  • Interesting- did you check it on V5-5500? It means 33MHz card handle much better O/C across agp FSB but in a 6000s you will not get too much advantage since it will lock up over 70MHz agp bus. At least in test I did using 6000s.


    Regards,

    Oscar.

    not tested yet

  • Hey Avenger You use the same EPoX mainboard as I do right? the EP-8K7A+ right?

    When I had my 6K with the EP-8K3A+ I never found much performance loss from 33Mhz & 66Mhz AGP bus to be very honest, also why the performance diff of a V5 5000 & 50000 PCI & 5500 AGP & 5500 PCI/PCI Mac wasn't much more than 5%, barely noticeable when playing glide games and such cases.

    T&L was never a thing for VSA-100's you'd need a decent Dual CPU setup 2x AthlonMP 2200+ or higher to emulate it, but that is like cheating the system and emulation is never as grand as the real deal, so I never participated in such useless tests.

    3D Analyze wasn't always ideal to work with either, using the Japanese made original Kiroutu-Shikou AGP2PCI Convertor PCI Card in mainboards with a Core 2 Quad of Dual 2/4Core Opteron system I never saw the point simple the faster CPU's would be doing the most of the work as where the V5 would just be the bottleneck.

    But anyways, the 3dfx Voodoo5 6000 AGP 128MB Rev.A 3700 I had was dead stable, no crashes at AGP x2 66Mhz FSB, in reality it was equivalent to AGP x1 or PCIx64 32Bit @ 66Mhz Bus.

    my old card named Green Leader


    Green Leader in my 3dfx test system at the time and the entire system it's self named Sveta 2002



    Good times.


  • my system runs rockstable :) normal in 4x FSAA sometimes in 8x FSAA (the whole day at LAN Partys)

    I´m not interestet at any benchmarks, Card runs all the time @ 166MHz clock Speed, and AGP 66MHz

    Based on a 8K5A3+ from Epox with an Athlon mobile 12x166MHz


    Why do you want to operate your card @ AGP 89MHz? u know electron migration and other effects?

    I see no reason to maltreat V5 6000 cards today outside there specification

  • Elektronenmigration? Bei 89 MHz? Never.

    Da müssten alte Computer oder UKW-Radios auch nicht mehr funktionieren.

    Die Fertigungsgröße der Karte ist so grob, die funktioniert sogar im Weltraum.

    „Ich bin nicht nachtragend, aber ich vergesse nichts.“