VOGONS


Reply 120 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t

Awesome chart. Too bad you don't have a system that will take it a bit faster. You could probably raise the top end to 166-200 Mhz and still have the bottom end be slow enough for WC1. Speaking of which, do you have any target games for this system?

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 121 of 194, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++

I wanted to cover 386 era games Dynamix games such as Deathtrack and Stellar 7, Sierra games, and Ultima 7. Perhaps some Wing Commander and Strike Commander as well. I'll see how fluid DN3D and Quake run with this, but I'm already contemplating an AM2 based DOS/Win98/WinXP build for more demanding titles. Still not sure what to use for a video card on that one, though. What PCIe card would be best for DOS compatibility?

This build could potentially be upgraded with a SS7 board as well, although I predict some overlap in the trend lines occurring with clock speeds too divergent. Really, an upper clock of 166mhz would be ideal, but my ASUS doesn't reliably do 83mhz FSB and 75mhz requires two jumpers to be flipped.

All hail the Great Capacitor Brand Finder

Reply 122 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

I wanted to cover 386 era games Dynamix games such as Deathtrack and Stellar 7, Sierra games, and Ultima 7. Perhaps some Wing Commander and Strike Commander as well. I'll see how fluid DN3D and Quake run with this, but I'm already contemplating an AM2 based DOS/Win98/WinXP build for more demanding titles. Still not sure what to use for a video card on that one, though. What PCIe card would be best for DOS compatibility?

This build could potentially be upgraded with a SS7 board as well, although I predict some overlap in the trend lines occurring with clock speeds too divergent. Really, an upper clock of 166mhz would be ideal, but my ASUS doesn't reliably do 83mhz FSB and 75mhz requires two jumpers to be flipped.

Ultima 7 should run good with the CCD+DCD switches. Wasn't Strike Commander a bit of a resource hog for 120Mhz? Or was that Pacific Strike? I'm curious how DN3D and Quake play. Looking forward to your report. BTW, which graphics card are you using?

I have zero experience with PCIe and DOS...

My high-end DOS gaming system uses a K6-2 550 and TNT2 M64 PCI. I don't have any Windows OSes on it (I'm not a fan of 9x since I can do 99.9% of that on XP).

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 123 of 194, by kithylin

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

I wanted to cover 386 era games Dynamix games such as Deathtrack and Stellar 7, Sierra games, and Ultima 7. Perhaps some Wing Commander and Strike Commander as well. I'll see how fluid DN3D and Quake run with this, but I'm already contemplating an AM2 based DOS/Win98/WinXP build for more demanding titles. Still not sure what to use for a video card on that one, though. What PCIe card would be best for DOS compatibility?

This build could potentially be upgraded with a SS7 board as well, although I predict some overlap in the trend lines occurring with clock speeds too divergent. Really, an upper clock of 166mhz would be ideal, but my ASUS doesn't reliably do 83mhz FSB and 75mhz requires two jumpers to be flipped.

I guess I still don't understand the point of all of this. Personally I've found instead of "trying to slow down fast systems" to just build something period-appropriate instead. Like build an actual 386 or 286 system for the older games and a really fast dos system based on AthlonXP or Pentium4 for the newer ones like quake/doom/duke3d/descent/blood, etc. It's usually much, much more compatible all around with zero issues.

Reply 124 of 194, by gdjacobs

User metadata
Rank l33t++
Rank
l33t++
kithylin wrote:

I guess I still don't understand the point of all of this. Personally I've found instead of "trying to slow down fast systems" to just build something period-appropriate instead. Like build an actual 386 or 286 system for the older games and a really fast dos system based on AthlonXP or Pentium4 for the newer ones like quake/doom/duke3d/descent/blood, etc. It's usually much, much more compatible all around with zero issues.

Space. I want to keep things compact.

All hail the Great Capacitor Brand Finder

Reply 125 of 194, by kithylin

User metadata
Rank l33t
Rank
l33t
gdjacobs wrote:

Space. I want to keep things compact.

Well what you're going to get in to and find eventually (I've seen it in a lot of older games) is some games physically don't run on the newer cpu architectures, no matter how slow you make them. Some older game engines will flat out crash at start up on P55C and newer for example. I don't remember off-hand which do and which don't but that is a "thing". At some point if you want to play certain older games you have to have original older hardware and there's no choice. Short of trying to emulate em.. but then they're more fun with nice older "real hardware" for the old midi music on nice physical sound cards.

Disabling caches and slowing down chips only gets you so far.

Reply 126 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t
kithylin wrote:
gdjacobs wrote:

I wanted to cover 386 era games Dynamix games such as Deathtrack and Stellar 7, Sierra games, and Ultima 7. Perhaps some Wing Commander and Strike Commander as well. I'll see how fluid DN3D and Quake run with this, but I'm already contemplating an AM2 based DOS/Win98/WinXP build for more demanding titles. Still not sure what to use for a video card on that one, though. What PCIe card would be best for DOS compatibility?

This build could potentially be upgraded with a SS7 board as well, although I predict some overlap in the trend lines occurring with clock speeds too divergent. Really, an upper clock of 166mhz would be ideal, but my ASUS doesn't reliably do 83mhz FSB and 75mhz requires two jumpers to be flipped.

I guess I still don't understand the point of all of this. Personally I've found instead of "trying to slow down fast systems" to just build something period-appropriate instead. Like build an actual 386 or 286 system for the older games and a really fast dos system based on AthlonXP or Pentium4 for the newer ones like quake/doom/duke3d/descent/blood, etc. It's usually much, much more compatible all around with zero issues.

No matter how you slice it, you still need two systems to cover the whole gamut. I settled on:
-486/66: slows down to 286 speeds with cache and turbo manipulation, so covers games from ~1985-1993.
-K6-2/550: slows down to slow 486, covers games from 1991-1997.
The overlap is nice since the K6-2 has the better General MIDI output for those few early 90s games that use GM.

The Test Registers are really useful for covering a wide range of games with one system. Like gdjacobs says, space.

kithylin wrote:

Disabling caches and slowing down chips only gets you so far.

Doesn't mean it's not useful. I think the only game I've encountered with the problems you state is WC1. It seems the faster the system (and thus more it's slowed down), the less consistent the framerate in WC1. I.e, when there's not enemies it's perfect, but with lots of enemies it's too slow, or vice versa. That's where, for me, it's better to play on the slowed down 486.

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 130 of 194, by Baoran

User metadata
Rank l33t
Rank
l33t

I have tried both my 100Mhz pentium and 120Mhz cpus with TR12 options and only option that seems to make much of a difference is L1DX which seems to do the same as "disable system cache" option in bios.

gdjacobs wrote:

The chip needs to support it. As far as I know, that's all you need.

Reply 131 of 194, by bluejeans

User metadata
Rank Member
Rank
Member

Odd "slow motion" timedemos in doom can be solved by turning off anything in bios involving shadowing, right? It's not jerky, it's just in slow motion and the realtics are all out of whack. On a pentium 2-350. Can't remember if it does it for my 166.

Reply 132 of 194, by dondiego

User metadata
Rank Member
Rank
Member

Timedemos have to run in slow motion on slow machines, it's curious that you consider normal fast motion on fast machines. There's no frame skipping to measure performance.

LZDoom, ZDoom32, ZDoom LE
RUDE (Doom)
Romero's Heresy II (Heretic)

Reply 133 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t

I think what he's saying is that the timedemo runs smoothly, just slowly (as if watching a movie in slow motion), when he expects the movement to be choppy and "steppy" like what we're used to seeing on a modern system when the framerate is in the single digits.

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 134 of 194, by bluejeans

User metadata
Rank Member
Rank
Member
clueless1 wrote:

I think what he's saying is that the timedemo runs smoothly, just slowly (as if watching a movie in slow motion), when he expects the movement to be choppy and "steppy" like what we're used to seeing on a modern system when the framerate is in the single digits.

That's correct. Doesn't look like timedemos on zdoom on a 486 that tell me they're running at 9fps.

Reply 135 of 194, by vvbee

User metadata
Rank Oldbie
Rank
Oldbie

I benchmarked against more or less pure floating point performance.

p/90                    100%
k6-2/300 no l1 39%
athlon xp/1.8 no l1 25%

486dx2/66 100%
k6-2/300 no l1 86%
k6-2/300 no l1/l2 26%
athlon xp/1.8 no l1 54%
athlon xp/1.8 no l1/l2 54%

386dx/40 100%
k6-2/300 no l1 1135%
k6-2/300 no l1/l2 352%
athlon xp/1.8 no l1 709%
athlon xp/1.8 no l1/l2 709%

Both k6 and athlon xp are too slow compared to the pentium/90. The athlon xp can't match a 66-level 486 even close, the k6 is approximately on par. Both are way too fast to be compared to the 386.

Reply 136 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t

@vvbee

What benchmark did you use for your floating point performance results? That's an interesting angle to look at it from.

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 138 of 194, by BSA Starfire

User metadata
Rank Oldbie
Rank
Oldbie

Added result's for Cyrix 5x86 100 on tomato 4DPS ver 2.1 motherboard, 32 mb FP RAM, number #9 S3 Virge 325.

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 139 of 194, by clueless1

User metadata
Rank l33t
Rank
l33t

@BSA Starfire - thank you. I fixed the Doom FPS column that wasn't filling and was write-protected, then resorted. Which reminds me, I have to retest my systems, as configurations have changed a bit.

edit: updated my 486DX2/66 results, adding disabled turbo results. Now I have the slowest system on the sheet! 😎

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