VOGONS

Common searches


DOSBox 0.74 CPU Benchmark

Topic actions

Reply 40 of 196, by kithylin

User metadata
Rank l33t
Rank
l33t
Tertz wrote:

Your 3rd picture for demo3 has correct results for demo1, demo2 and demo3. If these results were gotten on 2nd run of demonstrations then you may remember what you've done to get them. Maybe nothing, just some background process has finished.

Just to clear up that last bit, nope. Even though I did put AIDA64 cpu info in the final screenshots, I was closing it, running demos, then opening it, taking screenshot, closing it, running next demo, etc, etc. When running I had absoultely nothing running except the bare essentials for windows to actually be running. I even have themes (and aero) disabled on all of my computers, and all windows background scheduled tasks nuked as well.

Reply 42 of 196, by Standard Def Steve

User metadata
Rank Oldbie
Rank
Oldbie

Three different systems:

Pentium III-S running at 1575 MHz, Windows XP Pro SP3:
Demo 1: 6.3 fps
Demo 2: 7.4 fps
Demo 3: 6.3 fps

Socket 939 Opteron 185 running at 3133 MHz, Windows XP Pro SP3:
Demo 1: 28.0 fps
Demo 2: 30.9 fps
Demo 3: 27.6 fps

Core i7 4930K running at 4600 MHz, Windows 7 x64:
Demo 1: 62.8 fps
Demo 2: 70.9 fps
Demo 3: 62.4 fps

P6 chip. Triple the speed of the Pentium.
Tualatin: PIII-S @ 1628MHz | QDI Advance 12T | 2GB DDR-310 | 6800GT | X-Fi | 500GB HDD | 3DMark01: 14,059
Dothan: PM @ 2.9GHz | MSI Speedster FA4 | 2GB DDR2-580 | GTX 750Ti | X-Fi | 500GB SSD | 3DMark01: 43,190

Reply 43 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie

Thanks for your results, guys.

kithylin wrote:

Even though I did put AIDA64 cpu info in the final screenshots, I was closing it, running demos, then opening it, taking screenshot, closing it, running next demo, etc, etc.

Then it's not AIDA who's taken 10% of your CPU power, for example CPU-Z gets 7% fps on my system. But it's evidently that something has taken 10% or did not allow your CPU to use 100%.
34-35 fps is what should be gotten according to E8600 from same year. I don't know other ways to get stable +10% except frequency (you are sure it was same, but I'd recommend to check frequency only with CPU-Z as it's a specialized application and probably is more sensitive) or some software (OS, applications). Most probably it's software issues, wich concrete ones is a separate theme. As you may see, on 3rd picture you get 34 fps on demo1 but not 31, 39 fps in demo2 but not 36, - something has changed and has allowed your CPU to get correct results without 10% loss.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 44 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
Standard Def Steve wrote:

Pentium III-S running at 1575 MHz

Refinement of P3 model is needed. I see 4 variants.

Last edited by Tertz on 2015-05-18, 13:59. Edited 2 times in total.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 46 of 196, by kixs

User metadata
Rank l33t
Rank
l33t
Tertz wrote:
Standard Def Steve wrote:

Pentium III-S running at 1575 MHz

Exact P3 model is needed. I see 4 variants.

I think it's quite obvious that he used Tualatin P3 1400S overclocked to 1575.

Requests are also possible... /msg kixs

Reply 47 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
Qbix wrote:

maybe you accidently hit ctrl-f12 ? As max isn't the fastest DOSBox can go.

105% cycles gives me +3%, not 10%.

kixs wrote:

I think it's quite obvious that he used Tualatin P3 1400S overclocked to 1575.

While I don't think, as I had Intel CPU overclocked +50%.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 49 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
kixs wrote:

Fair enough...

Most probably 1400, but there can be variants: 1) hacked multiplier, 2) hypothetic super-motherboard allowing 210 x 7.5, 3) hypothetic motherboard with fractional FSB allowing something like 165.8 x 9.5. And it's not good to have the need to solve puzzles with FSB and multipliers.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 51 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
kixs wrote:

Usually the simplest explanation is correct and I'd go with 10.5x150 = 1575MHz for P-III 1400S 😉

Usually people don't overclock CPUs, don't use old hardware and don't write on retro forums. From local guys you may expect everything. 😀

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 52 of 196, by kithylin

User metadata
Rank l33t
Rank
l33t
Tertz wrote:
Thanks for your results, guys. […]
Show full quote

Thanks for your results, guys.

kithylin wrote:

Even though I did put AIDA64 cpu info in the final screenshots, I was closing it, running demos, then opening it, taking screenshot, closing it, running next demo, etc, etc.

Then it's not AIDA who's taken 10% of your CPU power, for example CPU-Z gets 7% fps on my system. But it's evidently that something has taken 10% or did not allow your CPU to use 100%.
34-35 fps is what should be gotten according to E8600 from same year. I don't know other ways to get stable +10% except frequency (you are sure it was same, but I'd recommend to check frequency only with CPU-Z as it's a specialized application and probably is more sensitive) or some software (OS, applications). Most probably it's software issues, wich concrete ones is a separate theme. As you may see, on 3rd picture you get 34 fps on demo1 but not 31, 39 fps in demo2 but not 36, - something has changed and has allowed your CPU to get correct results without 10% loss.

I'm not sure why you're not understanding me or I don't know what you're not getting here. When I ran the dosbox benchmark I had -NOTHING- running in the background to effect anything, no software, no monitoring, nothing but the bare minimum to keep windows running I even disabled networking temporarily. And yes, I can assure you 100% my cpu's speeds don't change, not even 1 Mhz in either direction. I can sit here watching cpu-z for hours, nothing changes.

There is no "magic 10% difference", what it is is what it is and that's it. And I was considering testing some of my other machines with this, but don't think I'll bother. You don't seem to believe anything I post anyway.

Tertz wrote:

Usually people don't overclock CPUs, don't use old hardware and don't write on retro forums. From local guys you may expect everything. 😀

Also just so you know, yes a lot of people overclock a lot of hardware. I overclock nearly everything I own, at least the things modern enough to find replacement chips easily and cheaply. Which is like AthlonXP and newer so far for me. And I do everything I can to prevent clock speed changes in all machines I own. Disabling speedstep, EIST, Cool`n`quiet (AMD), turbo modes, etc. All of my machines run with fixed clock speeds, even my big x58 i7 @ 4.4 ghz.

Things I don't overclock: my 386, my 486's, my socket7 machines, etc. Old stuff that's not easy to replace.

Reply 53 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
kithylin wrote:

You don't seem to believe anything I post anyway.

There was 10% loss of fps on 2nd measurement.
I belive to facts as fps and to conditions you've said when these fps were gotten. Except some doubts about frequency wich was better to be monitored by CPU-Z. I don't share your resume that 2nd results are correct and 1st are incorrect for demo1 (and others).
I'm finishing the argumentation about wich your results were correct as was said enough.

a lot of people overclock a lot of hardware

Most people don't overclock CPUs. And after Intel began to sabotage cooling of CPUs by placing a "bubble gum" under CPU's cap I think number of such people have rised.

And I do everything I can to prevent clock speed changes in all machines I own. Disabling speedstep, EIST, Cool`n`quiet (AMD), turbo modes, etc.

SpeedStep I find useful if it lowers power consumption and temperature when high load of CPU is not needed. Similar method is successfuly used for video cards. Turbo is useful for non-overclocked CPUs.
But overclocking is offtopic and for other themes.

Last edited by Tertz on 2015-05-19, 08:50. Edited 1 time in total.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 54 of 196, by Standard Def Steve

User metadata
Rank Oldbie
Rank
Oldbie
Tertz wrote:
Standard Def Steve wrote:

Pentium III-S running at 1575 MHz

Refinement of P3 model is needed. I see 4 variants.

Sorry, it's a 1400-S running at 10.5x150 for 1575MHz.

Here are a few older laptops running the benchmark:

Mobile Pentium 4 532 w/ HT, confirmed to be running at 3066 MHz during the benchmark (Windows XP Pro SP3):
Demo 1: 13.3 fps
Demo 2: 15.5 fps
Demo 3: 13.0 fps

Mobile Core 2 Duo T5600 confirmed to be running at 1833 MHz during the benchmark (Windows 7 x64 SP1):
Demo 1: 16.1 fps
Demo 2: 18.6 fps
Demo 3: 16.0 fps

That poor mobile Prescott actually did better than I thought it would.

P6 chip. Triple the speed of the Pentium.
Tualatin: PIII-S @ 1628MHz | QDI Advance 12T | 2GB DDR-310 | 6800GT | X-Fi | 500GB HDD | 3DMark01: 14,059
Dothan: PM @ 2.9GHz | MSI Speedster FA4 | 2GB DDR2-580 | GTX 750Ti | X-Fi | 500GB SSD | 3DMark01: 43,190

Reply 55 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie

Standard Def Steve

Thanks for new CPUs and the refinement.

That poor mobile Prescott actually did better than I thought it would.

Probably an improvement over Northwood and newer chipset.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 56 of 196, by LunarG

User metadata
Rank Oldbie
Rank
Oldbie

Intel Core i7 920 @ 3.78GHz (dosbox didn't seen to want to make it boost), 12GB RAM, Radeon R9 280X 3GB, Windows 7 64-bit: 39.8 - 44.7 - 39.6.
Seeing as I was in the middle of doing stuff, I didn't actually bother shutting down stuff just for this test, especially as the system is overkill for dosbox anyway. I had Pale Moon open with 4 tabs, skype, steam, anti-virus, software for Saitek X52 flight stick, text editor and a few other bits.

Intel Pentium III 1400S @ 1.4GHz, 512MB RAM, Matrox Parhelia, Windows XP SP3 32-bit: 5.2 - 6.1 - 5.1.
Seems I forgot to shut down ClaimWIN anti-virus on this system, but it doesn't utilize any noticeable amount of resources while idling anyway. Aside from that, only dosbox was running.
This system runs Adaptec ASC29160 Ultra160 SCSI host adapter with a 10.000rpm Seagate Cheetah Ultra320 HDD. But I doubt the storage solution makes any difference in performance for dosbox.

Not really sure what these tests are supposed to show though? I would imagine that very few people would bother putting together a dedicated system for dosbox, and if they did, wouldn't they just use cheap modern hardware rather than retro stuff? Any modern gaming PC would easily be able to run Quake at acceptable framerates in dosbox I would imagine.
Seeing as I prefer to use old hardware for old games, I guess I just don't know much about the dosbox community. 😀

WinXP : PIII 1.4GHz, 512MB RAM, 73GB SCSI HDD, Matrox Parhelia, SB Audigy 2.
Win98se : K6-3+ 500MHz, 256MB RAM, 80GB HDD, Matrox Millennium G400 MAX, Voodoo 2, SW1000XG.
DOS6.22 : Intel DX4, 64MB RAM, 1.6GB HDD, Diamond Stealth64 DRAM, GUS 1MB, SB16.

Reply 57 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
LunarG wrote:

Intel Core i7 920 @ 3.78GHz (dosbox didn't seen to want to make it boost)

Thanks for results.
About boost. As your system is overclocked, it may to have non-standard conditions for turning on turbo mode.

I didn't actually bother shutting down stuff just for this test, especially as the system is overkill for dosbox anyway.

This measurement loads ~100% of CPU, so some applications may noticeaply reduce fps. Antivirus, hardware monitors, active UL/DL processes, etc. You could lose 2-3 fps. But some reported getting more fps while having background apps, - the reason is a mystery still, probably some issues in OS with multi-core management.

Seems I forgot to shut down ClaimWIN anti-virus on this system

Compared to other Tualatin seems it could cost 5-7% of fps.

Not really sure what these tests are supposed to show though?

To know the progress of CPUs in emulation speed for post-1995 3D games.

Seeing as I prefer to use old hardware for old games, I guess I just don't know much about the dosbox community.

Quality and speed of emulation of pre-1995 games in DOSBox makes not much difference with old hardware now.

Last edited by Tertz on 2015-05-22, 18:04. Edited 2 times in total.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide

Reply 58 of 196, by LunarG

User metadata
Rank Oldbie
Rank
Oldbie

My i7 has all cpu features enabled, including boost. It boosts with 100% load and 8 threads in hyper pi, so not sure why it doesn't want to in dosbox. It's weird.

WinXP : PIII 1.4GHz, 512MB RAM, 73GB SCSI HDD, Matrox Parhelia, SB Audigy 2.
Win98se : K6-3+ 500MHz, 256MB RAM, 80GB HDD, Matrox Millennium G400 MAX, Voodoo 2, SW1000XG.
DOS6.22 : Intel DX4, 64MB RAM, 1.6GB HDD, Diamond Stealth64 DRAM, GUS 1MB, SB16.

Reply 59 of 196, by Tertz

User metadata
Rank Oldbie
Rank
Oldbie
LunarG wrote:

My i7 has all cpu features enabled, including boost. It boosts with 100% load and 8 threads in hyper pi, so not sure why it doesn't want to in dosbox. It's weird.

You may try to close or to hang concurent apps to one core and run dosbox on another, so you'd have less cores active and hence more possibilities for turbo. It's easier to experiment with SuperPI - another single-core application wich loads on 100%.

DOSBox CPU Benchmark
Yamaha YMF7x4 Guide