VOGONS


First post, by Comos

User metadata
Rank Newbie
Rank
Newbie

I have made this discovery already long years ago on my system,but due to lack of time I never got any reasonable explanation,so here is a short background.
When I succesfully upgraded my Asus TX97-X rev 1.02 mosfets for a higher load to be able to run AMD K6-III, I end that a setup 6x75Mhz was the highest stable seting,that I could run here (83FSB seems not supported by the PLL in this revision).Later on I moved to K6-III+ which I clocked with the same settings,but noticed in certain tools (MPXPlay example) which are able to show CPU usage,that show approx 3% higher CPU usage.The mobo has an updated BIOS to fully support K6 and the K6 DOS optimization tools which I run @ startup remain same setting,but Im not 100% sure if K6-III+ needed something extra right now.
Any idea what this small performance drop could be made of? One fact is that the K6-III+ is represented for 100FSB only altough it does run on lower FSB aswell.

Reply 2 of 7, by Comos

User metadata
Rank Newbie
Rank
Newbie
Intel486dx33 wrote:

I run my K6-3+450 @ 550mhz with 2.2v.
Runs stable with no issues.

Stability is not an issue,but a performance drop between K6-III+ and K6-III when clocked @ same speed.

Reply 3 of 7, by dionb

User metadata
Rank l33t++
Rank
l33t++

What performance drop?

Higher CPU usage means less waiting for I/O. Might be due to more efficient cache on the III+. What happens if you actually benchmark performance per second instead of CPU usage?

Reply 4 of 7, by derSammler

User metadata
Rank l33t
Rank
l33t

Stability is not an issue,but a performance drop between K6-III+ and K6-III when clocked @ same speed.

Are you aware that the K6-III+ was made for mobile devices, whereas the K6-III was made for desktops? It's to be expected that power-saving features causes slighly lower performance.

Reply 6 of 7, by Comos

User metadata
Rank Newbie
Rank
Newbie
derSammler wrote:

Stability is not an issue,but a performance drop between K6-III+ and K6-III when clocked @ same speed.

Are you aware that the K6-III+ was made for mobile devices, whereas the K6-III was made for desktops? It's to be expected that power-saving features causes slighly lower performance.

Yes,but the PowerNow! features set over the EPMR register are controlled over the SMM handler.To have the K6-III+ working on this board, the BIOS had to init this register upon reset into default,so under DOS it should not be active.To be really sure, I have to check that register (code some tool).

Reply 7 of 7, by Comos

User metadata
Rank Newbie
Rank
Newbie
dionb wrote:

What performance drop?

Higher CPU usage means less waiting for I/O. Might be due to more efficient cache on the III+. What happens if you actually benchmark performance per second instead of CPU usage?

Can do that,but wondering which bench tool use for that.Any example,that can give us exact numbers? (under DOS)