The official HWBOT OC Challenge June 2010 thread.

[Massman]

Just send a mail to the coder of the application. Hope he knows what's up.

 

This looks to be a problem of unsupported architecture. We had the same issue when the Gulftown appeared.

[aGeoM]

Just send a mail to the coder of the application. Hope he knows what's up.

 

This looks to be a problem of unsupported architecture. We had the same issue when the Gulftown appeared.

 

Hi, I already used 3 CPU's and 3 board's, Athon II X2 250, Sempron 140 and now Phenom II 965, boards are ASUS M2N32-Deluxe WE bios 5001, ASUS M3A79-T bios 1203, and ASUS Crosshair IV Formula bios 903.

 

Still playing NB and MLR and the last score is 1862.7 picture in a minute. LOL

 

 

Uploaded with ImageShack.us

 

[EDIT

 

http://cid-8828d15347fc782f.office.live.com/self.aspx/.Public/HWBOT/959l1730^_dc5501cc^_MaxxMem2.save

Edited June 10, 2010 by aGeoM

[knopflerbruce]

It could be something that isn't optimized for L3 cache in the code. Even so, a benchmark result isn't necessarily a good representation of real world performance.

 

I bet there are cases where HT actually harms performance as well, but it's not a bug:p

[chew*]

I'm on DDR3, and yes 52 is the lower I got booting at 9x on DDR2, and you can lower your MLR too, just configure lower NB in Bios and you will see. eg. 7x or 8x

 

Yep o know how to bring MRL down but if you rasie NB again with the tighter MRL = frozen screen

[aGeoM]

Ok, no tweaks settings from BIOS, NB @ 7x:

 

 

 

Tweaking NB, from 7x to 10:

 

 

No freeze, no reboot, just CPU-Tweaker to change NB, you have to open twice for new NB be accepted: open CPU-Tweaker -> change NB -> apply -> exit; then open-> chage some subtiming -> apply -> exit.

 

Run MaxMEM

[chew*]

Massman......there is more than meets the eye here still.

 

 

NO MRL.

 

 

MRL

 

 

 

BTW just figured it out........It's like a speed hack in 64 bit win 7 apparently.....

 

Cpu-z shows NB applies on the fly in windows........so does cpu twkr.....but in vista 32 score says otherwise.........will take a you tube of it

 

Also it's doesn't actually change NB in bios.......and now my OS thinks i'm at 2700 NB.

Edited June 11, 2010 by chew*

[aGeoM]

Just do the NB tweak, heres my last one NB @ 4700, when asked to reboot press [Cancel] and then reopen CPU Tweak.

 

 

MLR is at same 44 in the 3 cases, no change.

Edited June 11, 2010 by aGeoM

[chew*]

Yes i see that..........are you on LN2?

 

If No do you really think your at 4700 NB?

 

Do you really think it's applying?

 

no it isn't so therfore its an exploit of some form?

[aGeoM]

There is no exploit, just do what I say and you will see, only now testing an Black Edition I can go over 10x in NB w/ CPU Tweak.

 

Here's the last one, this time I set in BIOS NB@ 6x to force lower MLR (39) and then changed NB as I explain to you to 14x.

 

 

For some reason I don't know MaxMEM is acting as if I was at 4700, but results are consistent.

 

EDIT

 

Well, Why not go for extreme NB speed 30x...

 

Edited June 11, 2010 by aGeoM

[chew*]

the problem is GEO if your not understanding is your not actually at that NB speed you think you are.........Windows/or maxxmem however thinks you are .

 

Combined with the low MRL with the NB speed windows thinks you have it is impacting the bench.......call it an exploit or call it a cheat but your NB is not applying I am looking in bios.......it's not applied.........I'm in vista 32........although it says i'm at 2700 NB right now vista 32 is not tricked and performance is that of 800 NB which is what is set in bios

 

You would find that it would be impossible to run that tight an MRL at XXX NB speeds

 

Regardless of how you get it up whether that be an actual ability to raise via software.....which most boards can't do.

 

Believe me I am 1000% keen on the way MRL works and it does not work the way you think it does.

 

I suggest just my 2c that you clock your NB from bios and redo your results

 

I would dump win7 64 bit as well

 

As for me I'm all done with this comp till admins clean this mess up.......

Edited June 11, 2010 by chew*

[aGeoM]

the problem is GEO if your not understanding is your not actually at that NB speed you think you are.........Windows however thinks you are.

 

Combined with the low MRL with the NB speed windows thinks you have it is impacting the bench.......call it an exploit or call it a cheat but your NB is not applying I am looking in bios.......it's bot applied.........I'm in vista 32........although it says i'm at 2700 NB right now vista 32 is not tricked and performance is that of 800 NB which is what is set in bios

 

Now I know it's not possible my system is running at NB 4700, I booted many times at NB +3000 with sempron and Athlon, meanig 9x or 10x multis.

 

Both x86, x64 Win7s applied the NB setting in CPU-Tweaker, the first time I didn't realise what was triggering NB. Only with time I realised that setting NB first and then the timings, MaxMEM was performing faster. Found max NB, then was just playing with sub-timings, far from my mind to be cheating or something similar.

 

But you can duplicate my fiddings, just do the NB tweak, with multi-core, set affinity to one core.

[chew*]

Now I know it's not possible my system is running at NB 4700, I booted many times at NB +3000 with sempron and Athlon, meanig 9x or 10x multis.

 

Both x86, x64 Win7s applied the NB setting in CPU-Tweaker, the first time I didn't realise what was triggering NB. Only with time I realised that setting NB first and then the timings, MaxMEM was performing faster. Found max NB, then was just playing with sub-timings, far from my mind to be cheating or something similar.

 

But you can duplicate my fiddings, just do the NB tweak, with multi-core, set affinity to one core.

 

I could duplicate it but it's not legit....

 

I suggest you go back and do some legit results with booted NB multi now that it's quite clear that software was bugged.

 

If there was an actual program like AOD which may work on some boards on the fly.....you would find the inability to lower MRL just as if you booted it

 

Just some FYI, I can't duplicate what your doing in vista 32.......so admins may want to look into an OS designation for the remainder of this comp

Edited June 11, 2010 by chew*

[knopflerbruce]

Uhhh... I don't get this, is the benchmark just reading some parameters and determining the score based on that, or does it indeed measure some sort of memory performance?

[chew*]

Uhhh... I don't get this, is the benchmark just reading some parameters and determining the score based on that, or does it indeed measure some sort of memory performance?

 

I'm guessing a little of both......anything can be speedhacked though and thats preety much how the bench is reacting.

 

Gay bench. Needs to get canned IMHO.

[knopflerbruce]

Needs a fix, the bench itself looks cool:) But i bet there will be some sort of message tomorrow, but I'll leave that to the "big guys" when they wake up in 6-7 hours.

 

Jus for the record: alot of 3dmark versions are bugged as well, "bugged high, bugged low" - we've all heard that a few times. Time to remove them from the bot, so we're left with am3 + vantage perhaps?

[chew*]

Needs a fix, the bench itself looks cool:) But i bet there will be some sort of message tomorrow, but I'll leave that to the "big guys" when they wake up in 6-7 hours.

 

Jus for the record: alot of 3dmark versions are bugged as well, "bugged high, bugged low" - we've all heard that a few times. Time to remove them from the bot, so we're left with am3 + vantage perhaps?

 

I say we just bench the original 3d mark what was it 3dmaxx something or other?

[aGeoM]

I could duplicate it but it's not legit....

 

I suggest you go back and do some legit results with booted NB multi now that it's quite clear that software was bugged.

 

If there was an actual program like AOD which may work on some boards on the fly.....you would find the inability to lower MRL just as if you booted it

 

well.. chew, I'm not submitting results, any more, only here, with the intention of clarifying whats happening, CPU-Tweaker allow you to change NB freq, CPU-Z reads, MaxMEM acts as if NB is setting at that speed. You can see by looking to all my results that, in some I raise NB and CPU voltages bcz I was convicted that NB was running at that speed, and since I reached 10x limit I went HT ref route. AOD it's a fine tool, but if you change values there, like NB or HT and some in the memory page, it sets in BIOS to AUTO, insted os MANUAL, and we have to redo all the settings, HT Ref and some voltages...

 

About you Vista x86, you have to do the right steps to NB get read by MaxMEM, and I'm not sure but all AM2+ and AM3 board can change NB On the Fly w/ CPU-Tweak.

 

Any way be legit or not, using a software tool to change settings, I will remove my DDR2 and DDR3, for you please, not because I consider I was cheating or using exploids witch I was not, just used a tool in some cases required as validation (MEMSET/CPU-TWEAK) like any other tool (SPI Tweaker, A64Info, K10STAT, etc) you/we use to do your/our benchs.

[chew*]

well.. chew, I'm not submitting results, any more, only here, with the intention of clarifying whats happening, CPU-Tweaker allow you to change NB freq, CPU-Z reads, MaxMEM acts as if NB is setting at that speed. You can see by looking to all my results that, in some I raise NB and CPU voltages bcz I was convicted that NB was running at that speed, and since I reached 10x limit I went HT ref route. AOD it's a fine tool, but if you change values there, like NB or HT and some in the memory page, it sets in BIOS to AUTO, insted os MANUAL, and we have to redo all the settings, HT Ref and some voltages...

 

About you Vista x86, you have to do the right steps to NB get read by MaxMEM, and I'm not sure but all AM2+ and AM3 board can change NB On the Fly w/ CPU-Tweak.

 

Any way be legit or not, using a software tool to change settings, I will remove my DDR2 and DDR3, for you please, not because I consider I was cheating or using exploids witch I was not, just used a tool in some cases required as validation (MEMSET/CPU-TWEAK) like any other tool (SPI Tweaker, A64Info, K10STAT, etc) you/we use to do your/our benchs.

 

Yah not saying you actually knew what was happening, however it's very evident now.

I raised NB with cpu-TWKR to and cpu-z verified it....but it's not real thats where issue lies. Not your fault. You thought it was actually improving your score.

 

Now that you know however best to redo results

[aGeoM]

Yah not saying you actually knew what was happening, however it's very evident now.

I raised NB with cpu-TWKR to and cpu-z verified it....but it's not real thats where issue lies. Not your fault. You thought it was actually improving your score.

 

Now that you know however best to redo results

 

LOL... was really improving my scores, the score was obtained by the benchmark, some how it was reading all settings both CPU and memory and calculating from there.

 

I'm deleting all scores now. You are already in first.

[Massman]

Ouh, this looks bad. So basically, you're changing the NB clock in windows (which is impossible since change NBclk needs hard reboot) and all application read that it's set like this, so you get an unreal score.

 

Good catch!

[aGeoM]

Ouh, this looks bad. So basically, you're changing the NB clock in windows (which is impossible since change NBclk needs hard reboot) and all application read that it's set like this, so you get an unreal score.

 

Good catch!

 

Well, only with Phenom II I realise the "unreal" score before I used the M2N32, with that board NB was set by default to 8x max, so on CPU-Tweak I could change NB 8x to 10x by Apply and Cancel on Reboot message, after many trials, to level the NB default of the Sempron/Athlon II, as we know are locked to 10x.

 

The "exploid" as chew called, only begans when I was forced to proff that the results are legit, changed board to M3A79-T and later cpu to Phenom II, with unlocked NB first 11 then 12...14 and at last 30.

 

I removed all my scores now.

Edited June 12, 2010 by aGeoM

[Massman]

Just submit a few scores ... too bad I apparently killed ALL of my DDR2 boards.

[zoro]

how did you manage to do that?

[chew*]

Just submit a few scores ... too bad I apparently killed ALL of my DDR2 boards.

 

If you wan't I can make you feel better with my death toll

 

OCZ booster ( might be salvagable with some soldering work )

 

9800 PRO ( no clue, blew out a trace in AGP slot )

 

939 x2 3800 instadeath at 3.3vddr

 

DVD rom connected to system when card popped ( my guess is feedback through 12v rail )

 

Persistance pays though, finally got some ddr benching in.

[Massman]

how did you manage to do that?

 

I did it in the past because I overclock