yosarianilives
-
Posts
2264 -
Joined
-
Last visited
-
Days Won
9
Content Type
Profiles
Forums
Events
Blogs
Posts posted by yosarianilives
-
-
But I saw post team au was retiring from cc
-
1 hour ago, exaberries said:
eaven xtreme rule change is good, 6990 wouldn't be allowed because different core codename but other dual gpus that would work exist but are fairly rare
Not according to database
-
1
-
-
Okay here's my comments
AMD CPU - GOOD
Intel CPU
- ycruncher 1b - Can we allow xeons with same core count as consumer chips?
- 3dmark physics - F sku banned, although not explicitly. Overall seems fine and I like the idea, igp is an interesting challenge too
- cpu freq - stage is fine, just something to look out for there are ways to show all cores at a clock speed when it's just one core. Should be pretty obvious if done though
AMD GPU
- Heaven xtreme xfire - rules say use 2 videocards so it may seem that 6990 quadfire is allowed as it has cayman core and 590 was allowed in similar circumstance in team cup
- gpupi 1b - if server cards like mi25 not allowed then there are a total of 4 cards allowed for this stage
- Firestrike ultra - can we drop to extreme or normal firestrike, the vram requirement makes system ram oc have a large performance impact as seen last time we did fsu for country cup on dual gpu cards
Nvidia GPU - GOOD
-
Yeah I think there are still some concerns on some stages, I'll type a full list when I get to a desk today
-
47 minutes ago, saltycroissant said:
If we don't allow Titans, produo shouldn't be allowed either
Hands off my garbage, firestrike ultra uses A LOT of vram, we can't afford to be banning high end dual gpu cards. Otherwise it's literally just a system ram bench
Source: this isn't the first fsu dual gpu stage in a comp
-
7 hours ago, Leeghoofd said:
Who's Yos ?
Idk, probably some asshole
-
1
-
-
Guys people have already planned their vacation time around nov/Dec. As fun as it would be to cuck the aussies, just no. We can discuss dec/Jan for cc 2023
-
We've voted on this before, stays Nov Dec. But we could always do another poll, ultimate decision is up to Alby. Either way we should suggest as many stages as we can think of that we'd want otherwise we have no right to complain about bad stages after comp start.
-
if we did bclk I'd say just go with cpu bclk for simplicity, but apparently seby killed 2 ssds already from am4 bclk, not just the install either. so maybe skip this one lol
-
35 minutes ago, denvys5 said:
true ref. clock, or cpu ref clock?
If true ref., sounds bad, we have no way to distinguish it from cpu ref clock (cpuz shows cpu ref clock on AM4) and we will see a lot of 120mhz results, once everybody figures it out.
If cpu ref clock, boards with external bclk AND ref. clock target switch (for example, crosshair 6 and 7) will dominate the stage, which also doesn't sound great.W8 they're not linked? And I think we'd do it by chipset so at least c8 would be out cause x570 sux
-
10 hours ago, Speedy22 said:
H67= ?
H67 is H67, P67 is P67
-
As team cup 2022 draws to a close it's a good time to start proposing stages. I went through the team cup ones and kept the ones appropriate for cc as a starter but please comment with ideas for stages or risk getting laughed at when you wait until after comp starts to complain about stages. https://docs.google.com/document/d/18iCZWy45BrKM_on_K0JxEzWEizkrR6AX12JhDywwNd0/edit?usp=sharing
-
2
-
-
9 hours ago, TonyBombassolo said:
Is there a process or a history that explains why HWBot has an inconsistent application of allowing LOD/Tess manipulation on benchmark submissions? I can understand legacy benchmarks like 3DMark03, but I am talking about current, supported and active benchmarks.
Specifically, ALL of the Firestrike benchmarks have LOD/Tess explicitly allowed
- https://hwbot.org/rules?referenceId=5204 Fire Strike
- https://hwbot.org/rules?referenceId=5205 Fire Strike Extreme
- https://hwbot.org/rules?referenceId=5206 Fire Strike Ultra
For Time Spy, LOD/Tess manipulation is explicitly disallowed
- https://hwbot.org/rules?referenceId=5277 Time Spy
- https://hwbot.org/rules?referenceId=5179 Time Spy Extreme
Port Royal explicitly disallows LOD/Tess manipulation
- https://hwbot.org/rules?referenceId=5169 Port Royal
To me it seems like Fire Strike submissions allowing LOD/Tess are inconsistent with that allowance for other "current-world" benchmarks (in this case from the same vendor and the same suite of products) and Im failing to understand why that is. Typically HWBot is pretty consistent with benchmark rules, so this seemed odd to me. One of my favorite things are the new updates for Benchmate, because in general it removes "software skill" from the "overclocking skill" point system by allowing everyone to play on the proverbial same field. Any thoughts on why we would have inconsistent LOD/Tess rules on GPU benches, especially ones that give Global points?
It basically has to do with when systeminfo was able to actually detect it. Because there's lots of legacy scores where it could be used and not detected its still allowed as a tweak on those benches. By the release of Timespy and later system info had fixed this
-
On 10/7/2022 at 3:22 PM, Alastair_S1D said:
It says AM3 cpu's only. Is there no chance of using FM2+ based systems like an 860K? My FX's motherboard is long dead and my FM2+ system is the only DDR3 based AMD system I have knocking about.
The time to change stages is far far gone lol, comp ends on the 15th.
-
1
-
-
Lmao forced air is also convection. If we wanna go full science here you forgot evaporation which is how ln2 works. I wonder when we get 100% radiation based cooler ?
-
8 hours ago, Aleslammer said:
HDZ, C2
HDX, C3, Both BE's
Typically the stepping is at the end, gi is c2 and gm is C3. Hdz vs hdx typically is locked vs unlocked if you look across all the chips. Didn't look like 955 was different when I looked
-
1
-
-
5 hours ago, Aleslammer said:
AMD Phenom II X4 955 or BE I'm lost, seeing a lot of multi's way above 16 posted in the locked version. Know CPUz looks the same for both and I forget how x265 handled it.
HDZ955FBK4DGM Black Edition
HDZ955WBK4DGM Locked, 16 max multi CPU WorldI thought the locked version was hdx not hdz, but either way if cpuz detects the same then x265 would as well and need to manually select the right one based on ihs. If the multi is wrong it probably was a mistake on the side of the one subbing so report and it will be moved. I'd say 90% of people probably don't have the locked one.
-
It's hard to find out every unicorn before comp start. Although this one is kinda like a donkey someone taped a horn to and painted the word unicorn on the side. At least it was only like $40
-
I assume not?
-
5 hours ago, Leeghoofd said:
We allowed disabling before, only thing I can think off is to remove CPUZ it from the e.g. (P) versions
Roman has the idea to do only validations with all cores at the same frequency, now that will cause drama!
Definitely shouldn't do this, we joke that it's suggested just cause asus mad their lhe all core freq got beat by ln2
-
1
-
-
1 hour ago, Rauf said:
No WR points. Multi GPU 3D is irrelevant at this point (sadly). And 2D is just free points for best CPU in 32M, pifast and free points for the ones with access to server setups.
In general you are right, although I think emergence of ocable platforms that rival server like tr pro and w3175x makes wr pts actually about ocing and not borrowing clients cpu time. And for best results you still need ln2 which you won't run on borrowed system. Maybe wr pts are still bad idea for bot as a whole I just want to see people run ln2 and go hard on these big chips ?
-
3 hours ago, Splave said:
if you have multiple subs worth WR points do you get points for all of them if they are different cpu or core/card count?
Probably like any other pts where your best one counts. If you get best 8 core score on 12900k you can't get pts for 2nd best 8 core on 12700k. Same for wr, if you get best score overall with 64c and second with 32 you only get wr pts for the 64.
-
7 hours ago, Leeghoofd said:
Everything depends ofcouse which benchmark we selected for globals and which ones for the WR's or do we give WR points to each benchmark?
I think for sure no wr pts for single thread benches since their ranking already shows the wr, we'd get the same effect by just giving them more globals for top 10 scores or something.
For multithread benches I'm OK with all having wr though maybe some people will complain it makes expensive systems too op. Honestly I don't care about their opinions personally though lol.
For 3d idk, I wouldn't give it to legacy because most of them aren't gpu bound so single card is the fastest config. Similarly on newer stuff multi gpu is being dropped, already for greater than 2 way you need 10 series or older. Soon even 2 way seems to be going away so it may not be long before even on benches that aren't just about cpu oc that we see single card be the fastest. I think at minimum gpupi gets wr pts.
We should also make a new hardware category called accelerator cards so xeon phi doesn't take the wr pts from cpus on gpupi. Or maybe not, that could be interesting in its own way
-
Nice necro
-
1
-
The official Country Cup 2022 INTEL thread.
in HWBOT Competitions
Posted
Correct