Race Sim Central
Rec.Autos.Simulators Archive

anyone with cyrix/ibm 6x86 166 running N2?

16 posts by 11 people · 25 Mar 1997 - 31 Mar 1997

SBeshe 25 Mar 1997#360800
When you load n2 does it log out "pentium class machine"? My machine hangs when this is logged out. This happens when I run the cyrix optimizer programs (m1opt, 6x86opt, or cpuiden.exe) If i don't run these, n2 loads fine, but it doesn't display "pent.. cl.. mach..". My feeling is that when n2 detects pentium it loads a different instruction set. 486 vs. pentium. Running on asus ip55t2p4 rev 3.1 with 512 cache on board. Any suggestions?
Quoted text (11 lines)> When you load n2 does it log out "pentium class machine"? > My machine hangs when this is logged out. > This happens when I run the cyrix optimizer programs (m1opt, 6x86opt, or > cpuiden.exe) > If i don't run these, n2 loads fine, but it doesn't display "pent.. cl.. > mach..". > My feeling is that when n2 detects pentium it loads a different > instruction > set. 486 vs. pentium. > Running on asus ip55t2p4 rev 3.1 with 512 cache on board. > Any suggestions?
I run N2 with a p166+ and have not had any problems. What exactly do you mean by instruction set? Brian Johnson
Harley Pete 26 Mar 1997#360933
I have the same problem I happen to have the same motherboard and cpu that you have. The 686_p.exe program that i have seen on a lot of web pages does not solve the problem. They have apparently included some pentium instructions in the game that are not compatible with the cyrix 6x86. My question is since the program treats the 6x86 like a 486 how much is it costing me in performance?
Jim Sokolo 26 Mar 1997#360934
The program treats your chip as a 486, because that's what it basically is, a very, very fast 486. I don't think it's costing you much performance at all. With DMA based video cards (and to a lesser extent DMA for sound), you'll see your joystick reads be a little less exact (because the Cyrix doesn't support the Pentium RDTSC instruction). Other than that, there are no Pentium-specific features that are switched based on CPU_ID. (that I'm aware of.) ---Jim Sokoloff
AKH 26 Mar 1997#360935
The 'optimizer program' works fine for me (not sure if I get a performance increase) unless I use the CPUIDEN program. I'm running a Cyrix 166+. Nascar2 is the only program I have ran across that has a problem with the 'Pentium ID'. It will not run if I use the CPUIDEN program. In answer to the question about the program(s)...There are 'optimization' programs that are supposed to get more performance out of the Cyrix/IBM cpu's. These can be ran in your autoexec.bat or as needed. Atleast one of the programs has a switch that can be entered that will cause Win95 to ID the chip as a Pentium. CPUIDEN.exe is a program that will cause Win95 to report Pentium. For more info or to DL one of these check here: http://www.ionet.net/~rbdavis/ I'm not sure why Nascr2 would crash unless the Pentium version has optimized code for a Pentium. My best guess is that it does (Maybe Jim can enlighten us?) but I don't know what (if anything) that we loose with the Cyrix. Good Luck, Alan
Jim Sokolo 27 Mar 1997#361063
>In answer to the question about the program(s)...There are >'optimization' programs that are supposed to get more performance out of >the Cyrix/IBM cpu's. These can be ran in your autoexec.bat or as needed. >Atleast one of the programs has a switch that can be entered that will >cause Win95 to ID the chip as a Pentium. CPUIDEN.exe is a program that >will cause Win95 to report Pentium. [snip] >I'm not sure why Nascr2 would crash unless the Pentium version has >optimized code for a Pentium. My best guess is that it does (Maybe Jim >can enlighten us?) but I don't know what (if anything) that we loose >with the Cyrix. Well, N2 asks what CPU is installed, and if a Pentium is "detected", we feel free (and rightly so of course) to use Pentium instructions (the joystick read is really the only difference). A Cyrix chip is *NOT* a Pentium. If you install a program that causes the Cyrix chip to incorrectly report itself as a Pentium, you deserve for your programs to crash IMO. With a non-Pentium, you get a less accurate joystick routine (because of the lack of RDTSC on non-Pentiums). In any event, you are getting the full performance of your CPU. ---Jim Sokoloff
Jim Sokolo 27 Mar 1997#361064
> "a little less exact" that would be the understatment of the year. Well, true. As of this *instant* in time, I don't think it will be. If it is possible to fix, we will. We're trying a few things, but the lack of some Pentium instructions is obviously hampering... Obviously though, if we come up with a solution that works, we'll make it available. ---Jim Sokoloff
AKH 27 Mar 1997#361065
<SNIP>
Quoted text (20 lines)> >Atleast one of the programs has a switch that can be entered that > >will cause Win95 to ID the chip as a Pentium. CPUIDEN.exe is a > >program that will cause Win95 to report Pentium. > [snip] > >I'm not sure why Nascr2 would crash unless the Pentium version has > >optimized code for a Pentium. My best guess is that it does (Maybe Jim > >can enlighten us?) but I don't know what (if anything) that we loose > >with the Cyrix. > Well, N2 asks what CPU is installed, and if a Pentium is "detected", > we feel free (and rightly so of course) to use Pentium instructions > (the joystick read is really the only difference). > A Cyrix chip is *NOT* a Pentium. If you install a program that causes > the Cyrix chip to incorrectly report itself as a Pentium, you deserve > for your programs to crash IMO. > With a non-Pentium, you get a less accurate joystick routine (because > of the lack of RDTSC on non-Pentiums). In any event, you are getting > the full performance of your CPU. > ---Jim Sokoloff
I never said 'I' use the program that causes the system to read a Cyrix as a Pentium ;o) ...and actually for the reason you said. That said though, I have tried it just to see any differences (some people swear it makes a differnce for the better (I doubt it)). I saw where 'The Pits' http://www.theuspits.com has one of the ID programs available on the site for N2 so obviously somebody 'thinks' it'll do something for 'em! I do use an optimization program (without the 'Pentium ID') because it does seem to make a slight difference (perception and benchmarks). BUT, in your reply, you mentioned a different Joystick read routine for the Cyrix...Is that possibly in some way the source of the problem with N2 and Cyrix when running the Renddma version. I've yet to come across anyone who is able to run N2 Renddma with a Cyrix 6x86 that doesn't experience major probs with steering (T2). Nasrend doesn't have the problem with the steering. Renddma 'jerks' hard left. Is the Renddma/Cyrix problem something that can be 'fixed' (game card, tsr, patch, etc.) or is it something that probably can't be addressed due to the complexity of the program (and then the best answer is 'get a Pentium' ;o) ). ...of course then there's those AMD K series cpus... ;o) Thanks, Alan
tlym 28 Mar 1997#361177
Quoted text (15 lines)> >They have apparently included some pentium instructions in the game > >that are not compatible with the cyrix 6x86. > >My question is since the program treats the 6x86 like a 486 how much > >is it costing me in performance? > The program treats your chip as a 486, because that's what it > basically is, a very, very fast 486. > I don't think it's costing you much performance at all. With DMA based > video cards (and to a lesser extent DMA for sound), you'll see your > joystick reads be a little less exact (because the Cyrix doesn't > support the Pentium RDTSC instruction). Other than that, there are no > Pentium-specific features that are switched based on CPU_ID. (that I'm > aware of.) > ---Jim Sokoloff
"a little less exact" that would be the understatment of the year. Nascar11 is completly undrivable on my cyrix rendtion combo. Pappy tec support saaid this problem would be part of the patch,but by the sound of your post...will it or won't it? I can live without the rest of the patch,but not playing the rendition version kind of sucks,I bought my screammin 3d for the sole purpose of playing nas2 and Icr2, the latter still not working even whith the patch. I think it's time to kiss cyrix goodbye. Tim Yuhasz Mr.V
Jim Sokolo 28 Mar 1997#361178
[huge snip] I have a long, fairly technical description of why exactly the Cyrix (and AMD presumably) chips don't do well w.r.t joystick reads while using DMA extensively in this database a while back. (www.dejanews.com, search for Cyrix DMA RDTSC or something like that) We have confirmed with Cyrix that their chips prior to the (unreleasesd) M2 do not suppor the Pentium RDTSC instruction. We have been investigating alternative solutions for more accurate joystick routines for non-Pentium machines, but so far, we haven't found a solution that works for 100% of the cases. To be perfectly frank, this is not the highest priority problem facing Papyrus right now, as Cyrix and AMD combined make up less than 10% of the market. There are other problems which are receiving more attention, but this problem is not being ignored... ---Jim Sokoloff
DennyK 28 Mar 1997#361179
I have a cyrix 166+ and N2, Indy2, and F-1 all run fine. I upgraded to this from a Intel P133 there was an improvement in fps in all three of the games, as far as a problem with the joystick I never noticed it when I switched to the cyrix cpu all though I use a home made wheel and pedals. I did notice that n2 didn't recognize my cyrix as a pemtium, but who cares the bottom line is the cyrix works just fine and is faster for less money. One thing though I don't use a rendition video card. ( there is more to life then N2 and Indy2 ) and in my humble opinion that is the only thing I find the rendition cards good for. Denny M.
Quoted text (16 lines)> [huge snip] > I have a long, fairly technical description of why exactly the Cyrix > (and AMD presumably) chips don't do well w.r.t joystick reads while > using DMA extensively in this database a while back. > (www.dejanews.com, search for Cyrix DMA RDTSC or something like that) > We have confirmed with Cyrix that their chips prior to the > (unreleasesd) M2 do not suppor the Pentium RDTSC instruction. We have > been investigating alternative solutions for more accurate joystick > routines for non-Pentium machines, but so far, we haven't found a > solution that works for 100% of the cases. To be perfectly frank, this > is not the highest priority problem facing Papyrus right now, as Cyrix > and AMD combined make up less than 10% of the market. There are other > problems which are receiving more attention, but this problem is not > being ignored... > ---Jim Sokoloff
Rich Edg 28 Mar 1997#361180
>I have a long, fairly technical description of why exactly the Cyrix >(and AMD presumably) chips don't do well w.r.t joystick reads while >using DMA extensively in this database a while back. Just my $.02, but I have a P166+ and have had no problems running N2. I do not have any CPUID fooling routines running, but I have on occasion run one of the optimizing programs. ( 6x86opt ), again with no problems. I don't have a 3d card or a t2 ( aint givin up my Millenium for no SIERRA video card :-) ). Those of you having problems might want to drop back down to some similar configuration and try adding parts one at a time and ID the problem. I have no idea what the cyrix program from the pits does, but it obviously isnt really needed. Post your ideas and we'll come up with something.
Eric Busch 28 Mar 1997#361181
That's because the joystick problems occur with the 3D accelerated DMA version. --
Mike Radl 28 Mar 1997#361182
Quoted text (9 lines)>I have a cyrix 166+ and N2, Indy2, and F-1 all run fine. I upgraded to this >from a Intel P133 there was an improvement in fps in all three of the >games, as far as a problem with the joystick I never noticed it when I >switched to the cyrix cpu all though I use a home made wheel and pedals. I >did notice that n2 didn't recognize my cyrix as a pemtium, but who cares >the bottom line is the cyrix works just fine and is faster for less money. > One thing though I don't use a rendition video card. ( there is more to >life then N2 and Indy2 ) and in my humble opinion that is the only thing I >find the rendition cards good for.
Denny, AFAIK the only problem with using the Cyrix is when your running the Rendition DMA version. My limited understanding of the problem is that when the video board takes over the bus for DMA it throws off the timing used to calibrate the joystick. The Pentium has an instruction (RDTSC) that allows the new routine to get around the problem. The Cyrix doesn't have this instruction or an alternative therefore it's not 100% compatible and doesn't work very well. Fortunately the Cyrix does work fine with the standard N2 and Rendition nonDMA versions. Mike Radler Member Hawaii Ace League http://www.dithots.org/hal
AKH 28 Mar 1997#361183
To clarify things a bit... It is ONLY when using the combination of a Cyrix 6x86, Rendition vid card, and then running Nascar2's Renddma.bat that there is a controller problem (and I assume that is true for all using that combo). Nasrend, which is the 'regular' Rendition version, runs fine (even with Cyrix). The Renddma.bat version will have the fastest fps but can only be ran on systems that support DMA bus mastering (Depending on MB, Even some Intel systems can't run it). If you are using a Cyrix with ANY OTHER Vid card odds are great you WILL run as fine as your system will allow and you will not notice a control problem. You, of course, can't run EITHER Rendition version on a non Rendition video card. http://www.rendition.com As I understand Jim Sokoloff's reply, the Cyrix problem stems from a difference in the Pentium and Cyrix instruction set and how the program uses them. In summary: The Cyrix-Rendition-N2 control is fine for me in regular Rendition mode. The graphics ARE improved over the regular N2 graphics (non Rendition). If you are considering a Cyrix 6x86-Rendition combo be aware that you'll most likely NOT be able to use the fastest mode (Renddma) but that you'll be fine with the regular Rendition mode (Nasrend). You should see better graphics and fps with 'Nasrend'. Unfortunately, you won't be able to use the 'fastest fps''Renddma' version unless a Cyrix 'patch' comes down the pike...;o) ...hope that clears things up a bit... Alan
Quoted text (10 lines)> I have a cyrix 166+ and N2, Indy2, and F-1 all run fine. I upgraded to this > from a Intel P133 there was an improvement in fps in all three of the > games, as far as a problem with the joystick I never noticed it when I > switched to the cyrix cpu all though I use a home made wheel and pedals. I > did notice that n2 didn't recognize my cyrix as a pemtium, but who cares > the bottom line is the cyrix works just fine and is faster for less money. > One thing though I don't use a rendition video card. ( there is more to > life then N2 and Indy2 ) and in my humble opinion that is the only thing I > find the rendition cards good for. > Denny M.
<SNIP>
Alan...thanks...Chuck : To clarify things a bit... <good stuff snipped> : In summary: : The Cyrix-Rendition-N2 control is fine for me in regular Rendition mode. : The graphics ARE improved over the regular N2 graphics (non Rendition). : ...hope that clears things up a bit... : Alan

Sponsors

Affiliate Links

Sim Gaming Expo MOZA Racing Sim-Lab Fanatec TrakRacer Asetek Internet Privacy From NordVPN Dreamhost Capital One Credit Card Application Fanatical CDKeys Amazon Amazon UK Project Motor Racing iRacing Get your racing gloves, boots and more from Demon Tweeks. Enlist at Roberts Space Industries, developers of Star Citizen and Squadron 42


Back from the ashes since July, 2019. First created in 2001 with the merger of Legends Central (founded 1999) and simracing.dk.

A site by a sort of sim racer, for sim racers, about racing sims. News and information on both modern and historic sim racing software titles.

All products and licenses property of their respective owners.


Facebook Threads Bluesky Twitter Mastodon Instagram YouTube Tiktok Twitch Reddit Google News Apple News Subscribe via WordPress Flipboard eBay

Race Sim Central © 1999-2026 | Some links pay RSC a commission. Advertising does not equal endorsement.Cookie Policy