Step6 faster then Step5

[M.Beier]

Some time ago, after looking at Eva2000's 32M result, I began suspecting Step5 performed a bit worse than step6, so I decided to get it tested.

What I did was asking two members to run 32M, sadly first was a 965P user which I werent aware of, so last night I got the second result, same mobo as I, same mem settings.

The thing is, its primarily when overclocking the step6 benefits..

Here goes the test:

Many thanks too GOESA



In order too test results on your own setup run realtime, no other tweaks were made.

The results speaks for themselves, almots 2mins difference on 32M.

EDIT:
Post has been editied due not looking into the issue with forgetting too ask for 2*512 or 2*1GB, which obviously has an impact.. (please dont mind me being a bit angry with nobody willing to test this but just being b..chy..)

Many thanks too PcCI2iminal

- Futher I've made a rerun of my 32M in order for each round too be viewed, the rerun was less than 1 second different from the other one, which isnt a coincidence, please have a bit more trust in me, the settings are well thought off..
http://img241.imageshack.us/img241/2...inal32mte2.jpg (first run)

[Aphex_Tom_9]

wow

[linflas]

were the ram modules the same? I see your clocks are the same, but no reference to the actual ram used

[thelostrican]

i doubt it, some settings most be different, you are looking at a difference of 2 minutes....

[S&M]

I suspected that for a while now - this surely proves it ...
I get around the same time at pi32 with my step5 with 3.6ghz ....
So I think this is very intretsting ... !

[CedricFP]

What could possibly cause a discrepency of 2 entire minutes?

[lawrywild]

Both were run with the same Afinity + Priority settings? That's all I can think of that would cause a possible 2 mins difference...

[thelostrican]

same mobo crapier timings, as i could not remeber them all....
step 5
24-7 rig. no tweaks and a bunch of programs and crap in the background



edited for better pic

[mike]

Now to compare an A0 vs B2 would be real funny!

[Pt1t]

To have a good comparison, you have to test 32M on the same system and just switch the cpu.

[GordonFreemanLP]

Very interesting considerations....

Can you tell me what's the name of the program that shows core voltages on 2nd pic please??

[M.Beier]

#10, nope, just understandin' of hardware, made one mistake though, underestimated some ram-related issue...
I've received a few very different results since postin' which is hmm, an hour ago..

GOESA: A friend of mine, mem used: G.Skill 2GBHZ PC6400, the OS isnt in performance-mode, just classic skin choosen, no inet connection = less MS online applications... = a bit faster, however, very little difference.

However, still a half min difference



Now what I'd really like is someone with 2*512MB to go ahead and run as well.. BIOS settings for mine and GOESA: "Standart" in the section where you adjust voltage as well

EDIT:
THe program is Asus Probe2, I got alot of background applications running, doesnt hurt much when using DC and singlethreaded benchmark.

[mike]

the best way to see the differences would be to underclock - or at least stock clock the CPU and not run a memory intensive test. The higher the CPU is clocked at the smaller the actual time difference will be

[SlackeR]

Pt1t i think these results are pretty decent. Even more so after thelostrican filled us in with his time, which is in the Beier ballpark, even though it is unoptimized. Goes to show that optimisations are not the cause of the faster time.
Also, i know that Beier has been thinking about the problem - i actually know he chose not to ask someone because "he would not understand the "unoptimized" part"
I think the results are legit, though surprising.

[Pt1t]

Pt1t i think these results are pretty decent. Even more so after thelostrican filled us in with his time, which is in the Beier ballpark, even though it is unoptimized. Goes to show that optimisations are not the cause of the faster time.
Also, i know that Beier has been thinking about the problem - i actually know he chose not to ask someone because "he would not understand the "unoptimized" part"
I think the results are legit, though surprising.

No, i think there is maybe 10sec between B1-B2 for the 32M but not 2minutes.

[Kjaks]

Are you sure it isn't throttling? Try compare at lower speeds.

[G H Z]

Nothing to do with stepping, it's all in the setup.

[dinos22]

- Dont post if you do not have something decent to add to the conversation FUGGER

[death metal]

Step6 faster then Step5

Just some clarifications, to make such "statement", what is the confidence level that your findings is true "across the board", how do you define the "confidence" level and why can you say that a certain sample is good, or bad? Can you say that x% of all the CPU in XS is a good enough sample? Or can you say that x% of an estimated overall ES data collected across different sites then comparing them against x% of retail CPU is good enough basis to make such "conclusion" (if it is the conclusion)?

[GordonFreemanLP]

Hi.. Is anyone listening to my quest?

[dinos22]

Hi.. Is anyone listening to my quest?

PCProbe

Asus program

[M.Beier]

Just some clarifications, to make such "statement", what is the confidence level that your findings is true "across the board", how do you define the "confidence" level and why can you say that a certain sample is good, or bad? Can you say that x% of all the CPU in XS is a good enough sample? Or can you say that x% of an estimated overall ES data collected across different sites then comparing them against x% of retail CPU is good enough basis to make such "conclusion" (if it is the conclusion)?

clock for clock, and no I cant guarantee that all cpus will do 3600mhz

GordonFreemanLP:
Asus PC Probe2

Dinos22:
How dare you.

G H Z:
It isnt

Mike:
Any chance you could back this up, you got all the gear needed, right?

Hmm, Im not sure about the timerange, but think PcCI2iminal is getting a step6 as well very soon, and he can make same run with all the same settings, if that suits some of ya =)

[GordonFreemanLP]

Tnks a lot... :'( there are similar programs fon not-asus mobos??

[death metal]

clock for clock, and no I cant guarantee that all cpus will do 3600mhz

GordonFreemanLP:
Asus PC Probe2

Dinos22:
How dare you.

G H Z:
It isnt

Mike:
Any chance you could back this up, you got all the gear needed, right?

Hmm, Im not sure about the timerange, but think PcCI2iminal is getting a step6 as well very soon, and he can make same run with all the same settings, if that suits some of ya =)

Sorry for not being clear, I am not talking about the clock speed, but about your statement that All Step5 vs All Step6 = Your Header/Title.

If you are to ask me:
a) define a sample size, justify why such sample size is good for your "statement"
b) run the system in a completely controlled environment, that is, within your control and not some other system across the world where you may not know the exact settings (I am not saying people aren't telling the truth but people commit mistakes or understands things differently).
c) run the system where all other factor will not interfere, i.e. run them in stock settings at both speed/similar hard driver/etc. you get the picture.
d) record them in a systematic manner and not just a couple or more screenshots.
e) etc.

If anything, thelostrican just disproves the 2min difference you post.

[dinos22]

Dinos22:
How dare you.

lol