Let's benchmark our systems with caches disabled

Discussion about old PC hardware.

Re: Let's benchmark our systems with caches disabled

Postby Kamerat » 2016-6-28 @ 05:01

Added results for a Pentium II @ 100MHz with cache disabled, looks like 3DBENCH show's erroneous result at this speed, no way that looks like 57,5 fps.
User avatar
Kamerat
Oldbie
 
Posts: 848
Joined: 2014-3-14 @ 19:09
Location: Oslo, Norway

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-6-28 @ 11:07

Thanks for adding the results! Do you remember if you used 3dbench 1.0c or 1.0? Version 1.0 is supposed to scale better with slower cpus.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby Kamerat » 2016-6-29 @ 03:03

Updated the 3DBENCH result with the 1.0 version instead of 1.0c, looks more realistic now with 2,3 fps.
User avatar
Kamerat
Oldbie
 
Posts: 848
Joined: 2014-3-14 @ 19:09
Location: Oslo, Norway

Re: Let's benchmark our systems with caches disabled

Postby bluejeans » 2016-7-11 @ 23:43

I ran the pcp and 3d benches on a pentium 166 with pci voodoo banshee, then on an agp tnt2 with a p3-700 underclocked to 233. The scores were something like 5-10* higher, about what I expected due to AGP, so I put in a pci s3 trio and actually got a similar result on the 3d bench. Why is this? Surely the 3d bench wouldn't be taking advantage of the s3's meager 3d capabilities due to being run under real mode dos?
bluejeans
Member
 
Posts: 197
Joined: 2016-7-07 @ 20:35

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-12 @ 01:31

bluejeans wrote:I ran the pcp and 3d benches on a pentium 166 with pci voodoo banshee, then on an agp tnt2 with a p3-700 underclocked to 233. The scores were something like 5-10* higher, about what I expected due to AGP, so I put in a pci s3 trio and actually got a similar result on the 3d bench. Why is this? Surely the 3d bench wouldn't be taking advantage of the s3's meager 3d capabilities due to being run under real mode dos?

3DBench doesn't use 3D acceleration. Despite its name, it is a 2D performance benchmark, which the S3, Banshee and TNT2 are all very good at.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby bluejeans » 2016-7-12 @ 01:57

clueless1 wrote:
bluejeans wrote:I ran the pcp and 3d benches on a pentium 166 with pci voodoo banshee, then on an agp tnt2 with a p3-700 underclocked to 233. The scores were something like 5-10* higher, about what I expected due to AGP, so I put in a pci s3 trio and actually got a similar result on the 3d bench. Why is this? Surely the 3d bench wouldn't be taking advantage of the s3's meager 3d capabilities due to being run under real mode dos?

3DBench doesn't use 3D acceleration. Despite its name, it is a 2D performance benchmark, which the S3, Banshee and TNT2 are all very good at.


So the speed disparity with equal video cards must be due to having a p3 at 233, even with only a 77 mhz difference over the 166?
bluejeans
Member
 
Posts: 197
Joined: 2016-7-07 @ 20:35

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-12 @ 02:04

bluejeans wrote:So the speed disparity with equal video cards must be due to having a p3 at 233, even with only a 77 mhz difference over the 166?

Yeah. 77Mhz doesn't sound like a lot, but 233 is over 40% faster than 166. I'm not sure how a P3 and P1 compare clock-for-clock, that's another variable.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby bluejeans » 2016-7-12 @ 02:19

clueless1 wrote:
bluejeans wrote:So the speed disparity with equal video cards must be due to having a p3 at 233, even with only a 77 mhz difference over the 166?

Yeah. 77Mhz doesn't sound like a lot, but 233 is over 40% faster than 166. I'm not sure how a P3 and P1 compare clock-for-clock, that's another variable.


I had an idea for artificially limiting cpu speed in windows, simply adjust the power management so that to processor state is a certain percentage of whatever speed you want. However I take it you've made this utility since most of what we're using it for will not run under xp and later? (xp being the first o.s I'm aware of that you could limit clock speed)
bluejeans
Member
 
Posts: 197
Joined: 2016-7-07 @ 20:35

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-12 @ 02:26

bluejeans wrote:
clueless1 wrote:
bluejeans wrote:So the speed disparity with equal video cards must be due to having a p3 at 233, even with only a 77 mhz difference over the 166?

Yeah. 77Mhz doesn't sound like a lot, but 233 is over 40% faster than 166. I'm not sure how a P3 and P1 compare clock-for-clock, that's another variable.


I had an idea for artificially limiting cpu speed in windows, simply adjust the power management so that to processor state is a certain percentage of whatever speed you want. However I take it you've made this utility since most of what we're using it for will not run under xp and later? (xp being the first o.s I'm aware of that you could limit clock speed)

I did not make Setmul. Forum member gerwin did. :) It is a compatible with DOS, 95, 98 and ME only. Here's more info on it:
viewtopic.php?t=38613
For my purposes, it is for a DOS PC, to slow the system down to play older speed sensitive games.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-31 @ 01:12

Added an Intel 486DX2-66. Used Setmul to disable L1 and BIOS to disable L2. Disabling L2 did not add significantly to the slowdown. With cache disabled, the CPU is performing like a 20Mhz 386.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby gdjacobs » 2016-7-31 @ 01:50

Interesting. Not much slower than a K6-2 on 100% cache slowdown.
User avatar
gdjacobs
l33t++
 
Posts: 5615
Joined: 2015-11-03 @ 05:51
Location: The Great White North

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-31 @ 01:59

Not documented on the chart, but the turbo switch drops speeds to the equivalent of a 486-18. Now to see if the turbo switch in conjunction with disabling caches can yield some other speed zones.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby PhilsComputerLab » 2016-7-31 @ 06:35

486 can be awesome for slowing down tricks. Lots of combinations, but every board implements the Turbo a little bit different, so your results may vary :)

Still, the 486 is extremely flexible.
User avatar
PhilsComputerLab
Hardware Mod
 
Posts: 6179
Joined: 2014-9-28 @ 03:33
Location: Western Australia

Re: Let's benchmark our systems with caches disabled

Postby matze79 » 2016-7-31 @ 08:12

When i turn off L2 Cache my K6-3 hangs at the BIOS Screen showing Summary (Floppy Drives, Memorysize, Ports..)
I think this is because of l3 Cache. :O
https://www.dosreloaded.de - The German Retro DOS PC Community
https://www.retroianer.de
matze79
Oldbie
 
Posts: 1154
Joined: 2014-12-12 @ 14:25
Location: Germany, Frankonia

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-7-31 @ 11:46

matze79 wrote:When i turn off L2 Cache my K6-3 hangs at the BIOS Screen showing Summary (Floppy Drives, Memorysize, Ports..)
I think this is because of l3 Cache. :O

Do you have BIOS timings tweaked? I noticed on my 486, if I got really aggressive with memory and cache timings, the system would work at full speed but would lock up when I disabled cache! I expected the opposite. LOL. At any rate, I don't know if this has anything to do with your K6-3, but it's worth a look to set BIOS timings to default and try again?

Thanks!
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby bluejeans » 2016-11-11 @ 13:40

On my p2-350 underclocked to 233, if I run topbench it wildly switches between saying it's a p200 and a k5 116, with one cache on and one off. What does this mean?
bluejeans
Member
 
Posts: 197
Joined: 2016-7-07 @ 20:35

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-11-11 @ 13:56

bluejeans wrote:On my p2-350 underclocked to 233, if I run topbench it wildly switches between saying it's a p200 and a k5 116, with one cache on and one off. What does this mean?

Some older DOS benchmarks get really confused when you do things like disable caches, underclock, or run them on CPUs much newer than the benchmark. I remember topbench being one of those benchmarks.

I wouldn't worry about it, just run a range of benchmarks and focus more on the ones that make sense. I put more weight on real game benches, like Wolf3D, Doom and Quake.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby bluejeans » 2016-11-18 @ 02:38

My doom realtics are varying wildly from the equivalent systems on the list, and iirc even with the same cpu with no cache disabled. Topbench reckons the cpu is the same speed though. Could it be my fsb or ram speed/amount/type? The 3dbench was within 15% or so of a "real" system.
bluejeans
Member
 
Posts: 197
Joined: 2016-7-07 @ 20:35

Re: Let's benchmark our systems with caches disabled

Postby clueless1 » 2016-11-18 @ 14:23

bluejeans wrote:My doom realtics are varying wildly from the equivalent systems on the list, and iirc even with the same cpu with no cache disabled. Topbench reckons the cpu is the same speed though. Could it be my fsb or ram speed/amount/type? The 3dbench was within 15% or so of a "real" system.

What are the rest of your system specs? Motherboard, graphics card, cpu, motherboard cache, etc. Also post your scores with cache enable and disabled.
The more I learn, the more I realize how much I don't know.
Let's benchmark our systems with cache disabled
DOS PCI Graphics Card Benchmarks
User avatar
clueless1
l33t
 
Posts: 3654
Joined: 2015-12-22 @ 17:43
Location: Midwest US

Re: Let's benchmark our systems with caches disabled

Postby James-F » 2016-11-18 @ 15:01

Each run is randomly different?
Some corruption creating a random bottle neck in performance.
Try different RAM and make sure you don't go above the cacheable amount of the Motherboard.
If it doesn't solve the problem try changing the CPU.

As you disable the Caches the CPU relies more on the RAM each level of cache you disable.
User avatar
James-F
Oldbie
 
Posts: 1431
Joined: 2015-11-30 @ 04:10

PreviousNext

Return to General Old Hardware

Who is online

Users browsing this forum: appiah4, BinaryDemon, The Serpent Rider, Tiido and 13 guests