Old School is Best School Season 3 Round 2 Discussion

[websmile]

To start the support thread, I received feedback that Realbench can give trouble to some contestants.

Please bench with internet connection active bencause save file does not work, you have to submit directly after benchmark. Would also like to have feedback on open CL and compability issue, I know geforce 7 works as I tested it on new gen asrock board, but if you see problems, this is right place people can share experience on this

[TAGG]

Works fine for me, but its really hard to find a spot where the CPU is stable... Especially bad if the benchmark crashes during the last test, after what feels like an hour of running

[Mr.Scott]

Points are not being totaled correctly on the overview page.

[Strunkenbold]

S939 Opterons allowed? Seems like we already have some entries with those. (I hope you say yes, if so please add a line about it to the limitations)

[websmile]

Yes, Opterons are allowed for this special competition

[TASOS]

S939 Opterons allowed? Seems like we already have some entries with those. (I hope you say yes, if so please add a line about it to the limitations)

 

Yes

 

We had agreed on this , in the preparation thread

http://forum.hwbot.org/showthread.php?t=160802

[TASOS]

We have to make clear now , that we are at the beggining of this round , about the wallpaper beeing used.

 

Should we have drama in case the little R2 logo (at the left side) doesnt show on the validation screenshot ???

 

Except one or two , most of the results till now , have this problem.

 

How will the moderators handle this ?

[websmile]

I cannot even see the r2 at full scale without maginfying glass, so I don´t care^^

[Mr.Scott]

I cannot even see the r2 at full scale without maginfying glass, so I don´t care^^

 

I didn't even know it was there. lol

[TASOS]

I cannot even see the r2 at full scale without maginfying glass, so I don´t care^^

 

I didn't even know it was there. lol

 

 

 

I guessed , that many people didn't know either.

That's why i made this post.

 

 

Round 1 , wallpaper had a small R1 at the top left corner.

 

Now the R2 , is still at the left side , down lower.

 

 

Anyway

I would something like a hwbot staff formal statement about this wallpaper issue , just in case someone reports a result , for not using the official wallpaper for the exact round.

 

Someone could assume , that there are other members already benching later round stages and show them on-time.

 

Because if the "R" marking doesnt count , i can start benching for whatever round and stage i wont , whenever i like.

 

Then i can crop the screenshot , or even change the system date , to match the stage date.

 

Am i wrong ?

 

./

[lanbonden]

To start the support thread, I received feedback that Realbench can give trouble to some contestants.

Please bench with internet connection active bencause save file does not work, you have to submit directly after benchmark. Would also like to have feedback on open CL and compability issue, I know geforce 7 works as I tested it on new gen asrock board, but if you see problems, this is right place people can share experience on this

 

There seems to be some problem with nvidia 7 series cards (havent tested any other series) that they always award 14726 points in OpenCL. All submitet results so far got that score and Ive recived it with both a 7900gtx and a 7200gt

 

Not to mention that those DFI boards are known for being grumpy with 4GB ram

 

Ohh and good thing you mentioned that save file problem.. I didnt check how to submit so only saved a screenshoot.. glad there is plenty of time to rerun.

 

Question about OpenCL error, is it ok with a run where we get a score but this error for OpenCL instead of it showing ksample/sec? Ive had it all my runs and I see that 2 of the 3 submited scores got it as well.

Edited January 20, 2017 by lanbonden

[websmile]

@TASOS I will not remove any submission because of r1 background used. I will propose a change at background, maybe r3 bigger for next round, but it was nearly impossible to see that 1 and 2 background had difference, I will not blame ocers for this. @lanbonden as long as I get a point status for the test and the general result, this is ok for me. With the point cap at open cl, this might be system related or open cl, I can assure you that on new gen with 7800gtx I scored higher at this test and results vary

[Aleslammer]

With no OpenCL support looks like 14726 is it. Tried 1, 2 & 3 cards as will as 2.43 & 2.44, same score for OpenCL. Best score was with a 7950 X2 for the same clocks on the CPU & RAM but based on 1 run could be nothing more than back ground noise.

[TASOS]

There seems to be some problem with nvidia 7 series cards (havent tested any other series) that they always award 14726 points in OpenCL. All submitet results so far got that score and Ive recived it with both a 7900gtx and a 7200gt

 

 

Dont loose your time testing.

 

No Cuda support , no open cl.

Period

 

We needed nvidia 8 series and up , for this.

[skulstation]

ar we alowed to run older 3dmark versions?

[TASOS]

ar we alowed to run older 3dmark versions?

 

From what i know , things are clear about the rules

HWBOT 3DMark Ice Storm Benchmark Validation Regulations

 

All versions except v1.0.0

[Alpi]

"Multi-GPU allowed" It will be a big boost for.... I mena err.... It will be good for... For the rig pic !!!

[Dead Things]

Sorry for the noob question. What's the cure for "Program files modified or corrupted" error during Realbench run?

 

edit - Mybad, was still in diag mode. All good now. Thx!

 

edit2 - Scratch that, error still occurring...

Edited January 24, 2017 by Dead Things

[Bones]

Sorry for the noob question. What's the cure for "Program files modified or corrupted" error during Realbench run?

 

edit - Mybad, was still in diag mode. All good now. Thx!

 

edit2 - Scratch that, error still occurring...

 

If you disabled some services in the OS from starting that will cause it, these are mainly web related services - I'm not exactly sure which ones they are but I've ran into this after disabling some and re-enabling them solved it. If you slimmed down the install then you may have to run a standard OS install until you figure out exactly what services it's wanting and retweak the OS copy you are using for this bench.

 

I ran into this not more than an hour ago and that's how I figured it out, at least what I had causing it - I"m just not sure ATM which ones it is specifically but I will figure it out eventually.

 

EDIT:

As you said, scratch that last.

I just did a complete reinstall of the OS without changing anything in the install, GPU drivers installed and all was the same as the last time it ran fine yet it threw the error this time around. It's almost like either 939 just has a hard time with it or something - Really hard to say.

Edited January 24, 2017 by Bones

[TASOS]

"Houston , we have a setup down"

DFI NF4 SLI + Opteron 180

 

I've "lost" my current setup , during Realbench.

 

I've left it run the bench , cause it takes some time ... and after an hour , when i got back to check , i found it on a black screen with blinking cursor.

 

Shut down and new power up.

System keeps restarting all over again and again.

 

Clear cmos didnt help , neither one mem module only did.

 

I have a bad feeling that it needs more than basic troubleshooting.

[Dead Things]

EDIT:

As you said, scratch that last.

I just did a complete reinstall of the OS without changing anything in the install, GPU drivers installed and all was the same as the last time it ran fine yet it threw the error this time around. It's almost like either 939 just has a hard time with it or something - Really hard to say.

 

It appears something goes sideways when copying over the RealBench files to the bench machine. My solution was to copy over the zip file and then extract it locally. Working now.

[Alpi]

Oh Mate, I really sorry !

Do You know is it cpu or board ?

I wrote this yesterday to Bones :

 

"Good luck to You as well ! I wish all of us have good time, strong runs. Anyway, don't You think these benchmarks are anything but not for s939. Poor little chips. Like we doing linx on them. I hope no one will have dead chip till the end. Pay attention on temps ! So, up for the competition now !! "

 

I can't say enough time that how grateful I am to Hwbot staff (and the community also ) organizes these nice competitions for us so but also feel bad when any have loss of some irreplaceable old stuff.

Ok, it can be happen anytime all of us have accepted it I think but maybe we shouldn't forcing this.

[Dead Things]

DFI NF4 SLI + Opteron 180

 

I've thought my Ultra-D was dead a few times, but it keeps bouncing back. It's extremely sensitive to pressure around the CPU socket, so try to alleviate that as much as possible. Maybe try introducing some brief and slight flex to the PCB. I also find sometimes if I just give it a day off, it's good to go again. G'luck!

[Mr.Scott]

Overview board still not totaling correctly.

[Bones]

It appears something goes sideways when copying over the RealBench files to the bench machine. My solution was to copy over the zip file and then extract it locally. Working now.

 

Saw that happen too with file extraction from the zip file.

 

Extracted the folder on my DD and sent it to a flashdrive, took the drive out to the shack and copied it to the bench setup.

ATM it's working again so I guess I've finally got it sorted.... Until next time of course.