_mat_
-
Posts
1003 -
Joined
-
Last visited
-
Days Won
41
Content Type
Profiles
Forums
Events
Blogs
Posts posted by _mat_
-
-
16 hours ago, Leeghoofd said:
Just got the word in that XTU 2.0 beta is up for grabs, maybe include it in the challengers for thorough testing by the community???
Can't find anything online. Any chance I can get a hand on this?
-
BenchMate prior to 0.9 didn't support Windows 7 well enough to be usable. Please download the latest version and try again: https://benchmate.org
The next version will improve support for Windows 7, especially the 32 bit version.
-
1
-
-
This is not a validated score that has been uploaded with BenchMate. After you get the score like you did in the screen, you have to press "Save Result" in BenchMate. The BenchMate main dialog will disappear and the result dialog shows. Take the screenshot with the button on the result dialog (or F5). Press "Submit Score" in BenchMate to upload the score to HWBOT.
I will make a video soon, it's long overdue.
-
2 hours ago, zesen said:
Once used the program on windows 7. It did not work, it is not clear what kind of error, the logs are empty. Why it is impossible to wrap "try catch" and display everything to the user or to the error log
Is this about BenchMate, GPUPI or Geekbench 5?
-
1 hour ago, MrGenius said:
I presume that means fixing GPUPI 3.3.3. Since it is supported by GB. Yet doesn't work with MANY early processors. Like anything IVB or earlier...so far as I can tell. Which I know you know. But don't seem to give a rat's ass about. For some reason...
GPUPI only works if OpenCL works on the platform/OS and there is support for double precision. There is also a (carefully crafted, but slower) Legacy Version of GPUPI 3.3, that will be necessary on any systems, that can't install OpenCL 2.0.
I also have multiple alpha versions of GPUPI 4 ready, that completely removes the necessity of OpenCL and will run on nearly every x86 system I can think of. But it's not finished yet due to my work on BenchMate, which has much more value for the future of this community and benchmarking in general. I will finish it at some point of course, and I'm actually looking forward to it. It's relatively easy to code in comparison to anything that relates to BenchMate.
But I guess I don't give a rat's ass. <3
-
Doesn't work on me, I obviously love the pain.
-
2 hours ago, max1024 said:
The benchmark itself is made by the commander of experienced programmers, previous versions have worked well, I see no reason not to use the 5th version.
Good one! The "commander of experienced programmers" (whatever the hell that is) wasn't even aware of reliability problems when using certain timer functions in Windows 8 and 10. I also doubt that there will be any solution for Geekbench to mitigate these problems, although it was promised (weeks ago). The commander also publicly stated in the end that he doesn't care if his benchmark is used here. Which implicitly means that he doesn't care about any of our requests as well. So if we have got a problem with competitively benchmarking Geekbench, it will not be fixed.
2 hours ago, max1024 said:And with Benchmate, it’s not so simple, it doesn’t work for old systems starting with Windows 98 and higher, the stability of its work is also not always adequate. I pay tribute to the efforts and enthusiasm of _Mat_, but in my opinion it is impossible to keep up with all the innovations and new benchmarks coming out and updated with old ones, here need a team of developers IMHO.
Old operating systems as well as very old platforms have no chance of being fixed. They don't have the necessary technical requirements to meet the minimum standard, that I'm trying to create here. For example old platforms are missing timer sources with enough reliability and precision, while XP for instance can't be seriously secured against the many possible attacks. I don't see that as a problem. If you are benching old stuff on old OS versions, just bench like you did before.
Also, if you have any problems with the latest version of BenchMate, please post in the support thread and/or let me know with a bug report. It can't be perfect right out of the womb, that's not possible for such an extremely complex task! But in time it will be good enough to be used with all modern platforms without even thinking about it twice. It will just work, and that's a promise.
As for how many people need to be involved to make this work, I don't think that you have the experience to make an educated guess here. Not to forget that there is about no money at all in this for now, so having a team working on this without any funding is unrealistic (and not necessary at this point). Time will tell and my guess is that there might be opportunities with vendors and their benchmarks. If that happens it's not only the one benchmark with funding that will get improvements. All benchmarks will be improved, even old ones, because every new feature and every bugfix will happen for them as well. That's the beauty of BenchMate.
Oh, and in case I die, the source code goes to @Splave. It's already in my will, he will receive a key from my family to access the repositories. And even if that doesn't happen for one reason or another, you can always go back to use the old methods to bench.
-
1
-
2
-
1
-
-
Nowhere in this universe.
Please read the posts above, I will not answer questions about Geekbench anymore.
-
1
-
-
I will remove the categories "with BenchMate" with the release of BenchMate 0.10. Geekbench will no longer be supported due to reasons stated in the post above yours.
What happens to Geekbench 5 here on HWBOT is nothing I really care about. It's not a safe benchmark to compare results competitively, that's for sure.
-
Thank you for the clarification, Frederik.
MrGenius, you should consider not posting here when drunk.
-
1
-
-
13 minutes ago, Splave said:
maybe he will forget to rebuy the domain and someone that gives a shit can take it over
Domains are renewed automatically each year. If the debt can not be paid, the domain provider can give it back to tld provider. Happens mostly on bankruptcy cases.
Even then the domain will be flagged as deleted and enter a redemption phase of 30 days, where it can be bought back by the previous owner at any time (with extra fees).
An impeachment of Trump will be more likely.
-
1
-
-
It would be the recommended way to instantly start with forensics on the server and get behind this issue. If HWBOT was indeed hacked, it's not only HWBOT's security that suffers, but also our own. Our mail addresses might be out floating around in the open, passwords might be compromised.
To kill this off without any research by stating that Norton got compromised is really not the right way to respond to this.
-
4
-
-
Weird, it seems like the GPUPI result didn't get captured in BenchMate. If this happens again, could you send me a bug report right after the run? You can find it in BenchMate's menu.
-
2 hours ago, T.Rex said:
benchmark not found ?
Do you have a problem with BenchMate? If so, please explain it with more detail.
-
-
13 minutes ago, Mr.Scott said:
I would make that more of a priority.
Ok, I will have a look at it. ?
-
1
-
-
If you want texture and model verification, please use BenchMate 0.9.3. There will be another File Hash entry right below the executable. The number will always be the same unless something was tampered.
Edit: Please don't use any NVIDIA resolution trickery with BenchMate, Windows native functions won't detect it and BenchMate will capture a wrong screenshot. I will look into it though, it's on my list.
The real solution here is to make CPU-Z optional with BenchMate. It doesn't add anything that's not already on there (and much better because it's captured during the run). Unless SPD info is necessary, but that would be a single additional CPU-Z window (for now).
-
2
-
-
3 hours ago, KaRtA said:
I setup all my Ryzen on W7, why can't everyone else, cause like, I am super shit and rubbish with OS's??? Hell, I am doing it now for X570.
I've setup a Win 7 OS for old Intel platforms yesterday and it was nothing but trouble. Since this month a fresh SP1 install can't even check for Windows Updates without having the latest SHA256 update patches included. More of the same hassle when switching an existing W7 install from an old Intel platform to a new one. It doesn't work because of missing xHCI USB drivers. Same thing with Ryzen I guess, because last time I checked the mouse and keyboard didn't work and I couldn't log in. CINEBENCH R20 for example needs two additional patches to work on Windows 7 and doesn't work out-of-the-box either.
Ryzen 3000 on Windows 7 is another topic. Multiple people tried and it's not only not officially supported (fixes and speed improvements won't happen on Windows 7), but it seems to be not as stable and slower.
So in the end it's either a struggle to get it working on modern platforms or not the best choice anyways. And at some point in the future it might not be possible at all to go back to Windows 7. What happens then?
Also don't forget that we want new people to come into the community. Most of the new guys start with Windows 10 on a modern platform and if they can't participate with that and their first score gets removed instantly, they will never get into competitive overclocking on HWBOT.
-
4
-
-
5 hours ago, 0.0 said:
Your welcome and great job your doing with Benchmate, hope your not pulling too much hair out
. For me it's nice to be able to use some brain matter for other things than trying to remember something that happened 5 seconds ago (getting old)
Did notice on HM87 with x64 using W8.1 or Fedora the clock is forced to 32-bit by the OS as well as HM65 and W7, also same with W10 x64 and X99. Maybe there's a story there?
On another note has anyone approached the big competition organizers with regards to Benchmate? Should IMO help a lot if it becomes a requirement for them also.
Some hair turned gray already, so I guess I will go for the Sean Connery look when BenchMate has its first production release.
I have "fixed" this bug by not touching the 32 bit mode configuration for HPET's Timer0 on Windows 7 anymore, even if a 64 bit main counter is available. It seems to be a hardware bug I could not get rid of, even when I disable the interrupts and the main counter. Some internal flaw in the Intel implementation I guess, that was never revealed because the OS always sets up the HPET timer properly before the first rollover happens. I have no clue why this doesn't happen on Windows 10 though.
In any case, I've read over on OSR that a Windows kernel dev recommended to always use HPET in 32 bit mode, especially on 32 bit systems. There is also the problem of the main counter possibly not being latched correctly when it's 64 bit on a 32 bit platform, because reading 8 bytes is not atomic for 32 bit architecture. So the counter has to be read multiple times and checked for validity, which makes the already slow HPET even slower.
Of course a 32 bit implementation has to handle the rollover that happens about every 5 minutes, so there are more instructions involved to make it work. For 64 bit I don't do rollover checks because it needs 9 million days with a 24 MHz frequency until that happens. I would call that safe for non-server/workstation applications. So consider the HPET64 implementation as faster, which is another reason to use Windows 10 if you can.
-
On 9/13/2019 at 1:33 AM, MrGenius said:
Bug report sent.
Pentium 4 1.8GHz Willamette 256
ASUS P4P800-VM
Tried...
bcdedit /set debug off
bcdedit /set testsigning on
Still won't run.
I had a gook look at this and BenchMate 0.10 will work correctly on Windows 7 with 32 bit. It doesn't support kernel code integrity checks and there were other minor bugs that needed to be addressed. It should work with the next release and I hope you will have the time to take another look.
-
1
-
1
-
-
Yes, BenchMate 0.10 will NOT support Geekbench.
It will come with new benchmarks though like pifast. All integrations are in agreement with their respective developers. I will not throw days of work into the garbage anymore.
-
3
-
1
-
-
You can download BenchMate 0.9.3 and add the manually installed Geekbench to the launcher.
Just start BenchMate and click the + button below the application list. Select the executable of Geekbench (Geekbench 4.exe) and it will be added to the list.
-
1
-
-
The rules would stay exactly as they are now for everything that is not supported or when BenchMate is not used.
-
2
-
-
I think that and even more information should be available on a validation page for each result, just like CPU-Z has it.
So we have to define:
- What needs to be on the screen? This should be only the most important information to get a quick look of what's going on here.
- What needs to be available on the validation page? This should be the place to show everything. Maybe some information is withheld and only available for moderators.
Time to enforce bench mate for AMD in windows 10
in Old HWBOT Rules (pre 2021)
Posted · Edited by _mat_
The skew is too high, this looks more like the realtime priority bug where the validation couldn't keep up because of the overwhelming load. This is fixed with BenchMate 0.10 (coming very soon, if my fucking job lets me).
Was Geekbench run with realtime?