Jump to content
HWBOT Community Forums

HWBOT R5 Bugs/Features 2.0 - Report bugs or request new HWBOT features here


Recommended Posts

Posted (edited)

 

Nope, don't see a problem at all :P.

 

(it's fixed)

 

@Paco: How the hell do you even have an octa socket motherboard lol. We've discussed it before and decided that it wasn't worth the time to add it, because you're probably the only one with such a motherboard :). If we add a tab for octa sockets, it's going to be empty 99.99% of the time anyway. But I'll keep it in my mind, maybe we'll add it in the future but not right now.

Edited by Devroush
  • Replies 404
  • Created
  • Last Reply

Top Posters In This Topic

Posted

Thank you ever so much Devroush.

Actually I'm not the only one that has made 8-socket submissions in the past, (if they submitted them correctly is something else :P) I'm just the most persistent :D

Now I can get back to benching it, this will make it soooo much easier for me to keep track of my subs ;)

Again thank you very much & I'm glad I caught you in a good mood :ws:

 

Be well :celebration:

  • Crew
Posted (edited)

Bugs:

-Phenom achievement doesnt work

-socket 1156 pentiums and xeons doesnt count for the 1156 achievements

-some arrandale mobile cpu's are listed as socket 1156, should be 989

-clicking on cpu core in hardware browser results in a 404

 

Features:

-possibility to sort cpu benchmark results by socket, to find out which is the fastest of s478 for instance

edit:

-remove ucbench from hwbot as calculations seems to be buggy

-results are more coincidence as everything else, good example here (sse2 is slower than non-sse), many threads makes better results- there is no plausible explanation for this

Edited by Strunkenbold
Posted
Bugs:

-Phenom achievement doesnt work

-socket 1156 pentiums and xeons doesnt count for the 1156 achievements

-some arrandale mobile cpu's are listed as socket 1156, should be 989

-clicking on cpu core in hardware browser results in a 404

 

Features:

-possibility to sort cpu benchmark results by socket, to find out which is the fastest of s478 for instance

edit:

-remove ucbench from hwbot as calculations seems to be buggy

-results are more coincidence as everything else, good example here (sse2 is slower than non-sse), many threads makes better results- there is no plausible explanation for this

 

Added the bugs to our backlog. Some questions:

 

- Arrandale Mobile are 989? I'll forward that to the database manager

- The UCBench SSE2 'boost' is known - it's quite reproducable, a bit like CL in 3DM01.

- You can already search for benchmarks by socket: http://hwbot.org/search/submissions => socket is available under processor and videocard

  • Crew
Posted

- The UCBench SSE2 'boost' is known - it's quite reproducable, a bit like CL in 3DM01.

 

If you look closer to the result I posted, you see that the top score is done with 64k threads and non-sse. How is that to explain?

Also the "trick" with many threads is unexplainable. From wprime we know that creating that many threads on a single core cpu costs very much time. Why are the scores in UCBench getting better then? Tweaking ram or cpu for better results also dont make sense, cause you can just rerun the bench and hope for some fancy result popping up. (which will usually do and you beat much better configs)

  • Crew
Posted

- You can already search for benchmarks by socket: http://hwbot.org/search/submissions => socket is available under processor and videocard

 

What I wanted to say is, it would be more comfortable to have this implemented in the rankings. Like this:

 

Family: All | Intel | s478 | Pentium 4 | Pentium 4 3,0Ghz Prescott

 

Also search has some bugs:

-click on a result opens a new tab (which is correct) but also opens the result in the search tab (which is twice)

-searching for socket a gives no results (because s462 also exists)

-best am2 submission is a Phenom X2 555BE (??)

-searching for s478 mixes results with Socket P

 

and most annoying: It actually doesnt give you the best results! Look s2011 for instance

Posted (edited)
If you look closer to the result I posted, you see that the top score is done with 64k threads and non-sse. How is that to explain?

Also the "trick" with many threads is unexplainable. From wprime we know that creating that many threads on a single core cpu costs very much time. Why are the scores in UCBench getting better then? Tweaking ram or cpu for better results also dont make sense, cause you can just rerun the bench and hope for some fancy result popping up. (which will usually do and you beat much better configs)

 

1. The latest implementation of SSE on a cpu will be slower than previous one. A SSE4 cpu will perform better with SSE3, in your case, i386 > SSE2. Might not make sense but that is how it works (were talking about UCBench here just to be clear).

2. Apples vs Oranges. Wprime splits a mathematical operation across CPU Cores, UCbench generates dictionary keys. I think I had my own Dictionary generator somewhere, will try and upgrade the code for >8cores and see how it scales.

 

EDIT:

 

Just to show you what I meant at #2. Same operation split across 1/2/4 threads on a 2core cpu:

 

threadedexpander.png

Edited by GENiEBEN
Posted

Another idea...... a kinda history database of the #1 for each country (and each league) with the dates they were #1 for. A permanent memory of who was on top :)

Posted
Another idea...... a kinda history database of the #1 for each country (and each league) with the dates they were #1 for. A permanent memory of who was on top :)

 

Thirded. Would be awesome to see how long different titles are retained, and how often they change hands... If you don't follow daily, and thats the largest part of the audience, you miss most of the battles where top rankings are actively changing hands very frequently.

 

I'd come back to check this out regularly to catch up on where the toughest battles for first are and who is dominating. :)

Guest cowgut
Posted

Well its not just me then ok...maybe i can do better anyways

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


×
×
  • Create New...