Jump to content
HWBOT Community Forums

XTU will become pointless


Recommended Posts

4 hours ago, Leeghoofd said:

from the Intel guys:

Hi Albrecht,

the 9/9 version contains XTU Benchmark 2.0 beta.  Calling it beta so your team and Matthias can take a look before we finalize it.

 

https://downloadcenter.intel.com/download/24075/Intel-Extreme-Tuning-Utility-Intel-XTU-?product=66427

Great news! Good to see some members of the industry listening to our little HWBot.

Link to comment
Share on other sites

  • Crew

Is that a bad thing ?

I would prefer it to be one if possible a distinguishable "HWBOT" version for Win10 64bit only with a dedicated safefile that will only work on the BOT, no other versions possible to submit, to keep the scores in line and the mods can verify the settings

I hope Matt can help out with his expertise and maybe integrate it in BM, that would be cool

  • Thanks 1
Link to comment
Share on other sites

I only had a quick look, because I'm short on time currently. But I can already say that there are security fixes in there, so it's actually better.

The bench itself is better as well, it fully loads an 9900K during the whole test.

Regarding timers things don't look very bright. I'm pretty sure that there is no mitigation for Pre-Skylake in there, so timers will skew on older systems.

I will have an in-depth look soon.

So I guess the goal is to give XTU points again if it does alright?

Link to comment
Share on other sites

  • Crew

only if the new version adheres to the inputs you and Allen have given them Matt, We need a benchmark, that remains stable in output.

I'll test some heat scenarios, first on air and than put the CPU on the SS to see if I get the same output/ efficiency if this is still flawed, it will be a no go...

Maybe we can give it a spin in the PRO OC, so the big boys can mess with it and see if they can break, trick it... it that an option ?

  • Thanks 1
Link to comment
Share on other sites

1 hour ago, Leeghoofd said:

Maybe we can give it a spin in the PRO OC, so the big boys can mess with it and see if they can break, trick it... it that an option ?

In my opinion both is needed. A theoretical research to check what the benchmark does (or doesn't do) and a small competition to see if it scales well, there are no oddities under extreme conditions and to figure out if it's fun to bench.

  • Thanks 2
Link to comment
Share on other sites

18 hours ago, Leeghoofd said:

Is that a bad thing ?

I would prefer it to be one if possible a distinguishable "HWBOT" version for Win10 64bit only with a dedicated safefile that will only work on the BOT, no other versions possible to submit, to keep the scores in line and the mods can verify the settings

I hope Matt can help out with his expertise and maybe integrate it in BM, that would be cool

Just thought a wider user base would be wanted, then again I'd also love to see more done on the Linux side. Maybe I dream too much :)

As for timer vulnerability on 6.5.1.355

xtu2.thumb.png.0e4299b06da1a895583edb702844e296.png

On the one hand most benches seem to suffer from this as discussed before with Matt, on the other hand with it running with elevated privilege and access to HW it should be able to do better if it wanted to, or are we taking things too far?

Link to comment
Share on other sites

Very quiet here, I'm having deja vu from when these problems were first reported over 4 1/2 years ago. ?

I'll just add that it also installs on W7 but not W8.1. I made a small program to run the bench without XTU so can check the performance still on W8.1 which as I am interested more in the performance side than the points is fine.

PBench.png.665b194e19fab44eb7b42ee273c02161.png

Run at 4GHz without HT. Unfortunately my i7-6800k is a poor overclocker, not even making the 100% bin list on silicon lottery :( As a bonus only takes a couple of secs to run instead of the long 20 runs on XTU. But enough of that as it seems it may have been in vain. The beta XTU2 bench is something different and doesn't even use AVX2 but FWIW I didn't find anything other than the timer issue mentioned above.

Link to comment
Share on other sites

Great stuff! :)

I only had a late part of an evening again, but it didn't take very long to remove the obfuscation of XTU to have about 90% of the code readable and ready for debugging. I also removed the anti tampering check with switching a single IL instruction. What does that mean exactly? Well, I basically have my own custom XTU version and can make my own scores. ;)

That said, the new version is better when it comes to Anti Tampering. Just not by a lot. The problem here is C#. Like Java it's the wrong tool for the job.

@0.0 I guess your wrapper is for the XTU1 score only, right? The XTU2 benchmark option is implemented by calling native code from XtuBenchmark.dll and registering a callback to fetch the score. I didn't go into it in all its depth, but this is a lot harder than to wrap the p95bench.exe.

Edited by _mat_
Link to comment
Share on other sites

21 hours ago, _mat_ said:

 

@0.0 I guess your wrapper is for the XTU1 score only, right?

Correct. It runs with just the p95bench.exe file, no need for XTU to be running or even installed. XTU2 is at the point I don't want to put effort into it and perhaps beyond me to  do anything with which is a good thing I think. XTU1 is fun though, ran it on an old mobile core2 duo, seemed to go well even though the CPU TSC isn't non-stop. Even managed to find a 32-bit p95bench from XTU 6.4 that would work with Vista 32-bit. Perhaps as expected the results are a little less for the 32-bit bench, around 5 to 8%. Shame I don't have a Ryzen system to try.

So looks like XTU2 is headed in the right direction although not sure why AVX2 was dropped. If those kind of changes are going to be made might it not be better to bring AVX512 to the show?, not that I have anything with AVX512 myself :(

EDIT:

Looking at some 6800k XTU results, one stood out as too high. The "wall" for it pretty well sums it up

wall.thumb.png.e9175f6e8916bc8bb4be0021ca750ff4.png

Question is, how many people would be willing to do this. 1 in 10000, 1 in 10? But I guess it only takes 1 to spoil it unfortunately.

Edited by 0.0
Link to comment
Share on other sites

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...