skullbringer
-
Posts
57 -
Joined
-
Last visited
-
Days Won
2
Content Type
Profiles
Forums
Events
Blogs
Posts posted by skullbringer
-
-
2 minutes ago, sergmann said:
yes, leeghoofd just confirmed in discord. it does not matter when the bios was uploaded, there is no deadline.
might as well make a secret new bios and only upload it 5 minute before competition end. mods don't see any issue with that.
- 1
- 1
- 1
-
new AVX512-enabled bios for Z790 Apex, 0022: https://drive.google.com/file/d/1Kt6JhLbDr6a_qDeC1hWV5NeZSx19LVjZ/view?usp=share_link
-
how about ammending "each bios must have unique version number and date showing in cpu-z"?
otherwise 000 bios files with no release date cannot be tracked at all. 000 uploaded here might not be the 000 used in stage
- 3
- 1
-
manufacturer websites can change, hence we almost "have" to do it via this forum thread. but this then also implies, whatever bios is not posted here before stage start, is no go
e.g. currently no Asrock or Gigabyte Board is legal to be used.
@Leeghoofd please confirm
-
Here are some bios files for Z690 Unify-X
https://drive.google.com/file/d/1Kpmzriffr7l4qBjHyV5UHOxw0p7ubuF6/view?usp=share_link
https://drive.google.com/file/d/1Ko9Aozkw1ZcFi4GtDoei5bDoBt00LDEX/view?usp=share_link
- 1
-
Do all bios used in compo HAVE to be posted in THIS thread BEFORE stage start?
cause if so, currently only Z790 Apex would be legal to use
-
Good luck everybody, let's have a fair competition without rule changes midway through! ?
- 1
- 1
-
I want this board NOW!
- 1
-
DDR5 3500 MHz 1:1 is really impressive stuff!
-
such a beautiful board!
-
-
works on x3i too, tested with dell and patriot modules with "locked" renesas pmic p8911
- 1
-
28 minutes ago, Vertex_FR said:
mem set for 6000C30, but it doesnt train at all, got 0F directly and it need clear cmos to boot again.
retry button or direct to bios doesnt work, cmos on board too.
need to unplug psu and press cmos on IOwhich ucode do you have selected 0x9?
-
8 minutes ago, Leeghoofd said:
Correct analysis by Vertex
@skullbringer You already know where the problem lies, you want a fellow OCer to confirm "from a distance" it is your board? Would that make your board perform better?
Only way to get it done is either to RMA it, sell it and/or buy a new one... Continuing to run around in circles or starting a lively discussion doesn't get you anywhere.Even if these are labelled designed for OC there are so many variables that can make each retail product different from one another. For what it is worth, this is not a DDR5 thing only, it happens with every generation. Though due to the high frequency demand it stands out quicker...
k guess forums are not meant for discussions anymore then.
besides, before today I always got the statement: your cpu is shit or your settings are wrong. never once was I told my board could be faulty. now you say it's obvious my board is faulty.
thank you mr moderator for translating.
-
Pretending problems don't exist, doesn't help, I just want to get to the bottom of what I have been chasing for 6 months.
To be clear, I would be very happy if my incompetence would explain everything. Other then that I can only test my half a dozen CPUs, half a dozen hynix memory kits, half a dozen bios versions and when I still come to the same conclusion, consider the board to be at fault.
I would think that a lively discussion to get to the bottom of things is not something that would require the attention of mods, or am I not allowed to call myself potentially incompetent? LUL
-
1 minute ago, sergmann said:
I think you don’t know, but all Tachyon are tested for 7k XMP. If they can’t do it, it goes back.
If you think my board is prebinned, I must disappoint you, it’s same board that everyone else has ?
k cool, then I only need to get my hands on one of those to prove my board is RMA worthy... oh wait they don't exist.
I know it's not my CPU because I have tried 4 different i9 chips, the best of which boots 7650 dual-channel into Windows on the 2022 Apex and reliably posts 6933 on the UX. Meanwhile on the Tachyon with F6d I need 20-30 training attempts to post 6933 once. (2T in all cases)
I'm also fairly sure it's not my settings. I've tried BIOS X3c, X3i, F6d and your profiles. I've found the sweetspot for VDD2 and CPU VDDQ for every CPU individually and again, the best I can reliably train in 2T is 6800, but only on the F6d bios which sadly does not have RTL tuning support.
So honestly, do you think I am that incompetent, or do you think my board is faulty and I should try to RMA?
- 1
-
10 hours ago, sergmann said:
it`t not the Board or Bios, it`s your CPU or wrong settings.
Try my profile
k, then let's exchange boards. you send me one of yours and you get my retail sample
- 1
-
5 hours ago, Vertex_FR said:
Nice ! i also can do 6800C32 on retail bios. not possible with XOC bios.
Did you try the patriot kit on the Tachyon ?not yet, ux more consistent, apex higher clocks.
maybe when the tachyon gets a bios that can train 6800 on my board with tight rtl's we can talk LUL
-
hahaha nice, that's more like it! ;)
-
managed to do 6800 c32 1T stable once. when I try to reproduce, it either does not train or is unstable.
also my dimm slots seem to have vertical play when latches are closed and when I lightly touch a stick, it errors more or freezes completely. ask support?
very hit and miss training still, but when it works F6d clocks best for me
-
damn nice score, ln2 going stronk
- 1
-
-
1 hour ago, Leeghoofd said:
This particular mindset is what lead to all these rules/limitations at the Bot... Remember what was the inital rule ? I guess it was 5206Mhz.
People benched it and it turned out that XTU reads out 5.22. Now you guys bench 5.22 XTU but CPUZ reads out 5245Mhz... Try to explain it to me...Logic prevails, ever thought why do you think we ask CPUZ too and not only XTU benchmark window?
Who's really bending the rules like rubber? Oh it's not stated in the rules, so we can abuse it and clock up higher coz XTU doesn't detect it properly...
Nice try lads, insert coin and try again... Try to beat the systems, guess what the system always wins...
Now if you insist I can still add hwinfo too...
Every competition needs clear, enforceable rules, otherwise it's not a fair competition. Imagine football without a clear, straight, white goal line...
You say logic prevails, yet cpu-z only reports the clock at time of the screenshot and not of the benchmark run, so how is it useful for monitoring the clock of the cpu during the benchmark run?
XTU clock monitoring is completely fubar, as it sometimes reads several hundred MHz above or below the actual clock which HWinfo would report. I asked about this a few weeks back and never got a clear answer, other than "read the rules". Cool, so I did and followed them to the letter.
You can add "Hwinfo running in background during benchmark run", but how is that any better if you can manipulate it even easier, by setting the polling time to several minutes, stalling the process with low priority? My 5-year old, non-existant brother can figure that stuff out.
Isn't that the very reason why Benchmate exists? Stage 3 with Pyprime 2b via Benchmate is an amazingly fun discipline, also because there is no ambiguity around following the rules of the clock limit. That's how all the stages should ideally be. If you still want to use other benchmark software that for whatever reason can not work or is not allowed to work with Benchmate, then there will always be these kinda loopholes and people that try to use them to their advantage. That's also in the very nature of every competition.
Ideally you would even want to have a standardized "HWbot OS" image that is completely read only and allows for no tampering at all, so that actually only the hardware and its configuration matter. But until we get there, and Benchmate is already a big step in that direction, there will always be tweaks and greyzones that people will use to their advantage, like tesselation in 3DMark or whatever trickery that HWbot rules officially take into consideration by now.
Even though I don't find it nice to get talked down to like in the post I'm responding to, I'm not gonna take issue with that or make it personal.
However, what I do take issue with is the changing of rules during the course of a competition. It already happened during the last stage of GSkill Tweakers where suddenly another Benchmate version was legal to use, while before that I had spent several weeks to get the only previously legal version running on Win7. So I'm kinda getting PTSD now that you consider adding Hwinfo to the XTU stage and thus invalidating previous submissions...
To add some more constructive criticism, I think if there is ever the need for the change of rules during a competition, there should be a poll sent to all participants where they have a few days time to agree, object or withhold if they don't care. If noone objects until the deadline, the rule change is admitted. Other "sporting organizations" do it just like that, too.
Overclocking competitions are extremely fun, but they have to be fair. If loopholes in the rules are found, brushing them aside and calling for the "spirit of the rules" to be adhered isn't a solution. Instead those loopholes should be openly talked about and addressed in a transparent and fair way. Because only then can we grow as competitiors and as a community of enthusiasts.
-
geekbench 5 clock readout is broken if bclk is not 100, this is not news ;)
skullbringer - DDR5 SDRAM @ 5023MHz - 5023 MHz Memory Frequency
in Result Discussions
Posted