-
Posts
2268 -
Joined
-
Last visited
-
Days Won
9
Content Type
Profiles
Forums
Events
Blogs
Posts posted by yosarianilives
-
-
I agree with removing 25m, it's too fast, kinda like super pi 16k. Can we add the 2.5b preset? 1b is getting too fast on high core count, talking under 10s and it becomes more about optimizing thread spawning and less about oc. Dunno if 2.5b should have globals tho cause it uses over 16gb of ram iirc.
-
1 hour ago, unityofsaints said:
Top strap on that board is horrendous, I think even 1:2 might work better. Ultimately you should just get a Lanparty though
800 mhz strap clocks the same, had this on several boards. 1066 just sucks for performance cause it loose
-
If you edit can you change the category? Usually works for me
-
Need to try more boards, this one doesn't work with manual timings, even after a reflash
-
Lmao someone told me the ddr record couldn't be beaten without crazy es from time period. Clearly they are wrong, let's see a ddr and ddr4 wr this comp. Ddr3 may be possible but I doubt it
- 1
-
38 minutes ago, Jumper118 said:
Do you have to to lower the pll voltage like sandy bridge?
Never tried that lol
-
40 minutes ago, StingerYar said:
Is there a possibility to add rule like "No cores or HT disabling allowed" to HyperPi stage? Currently it just says "HyperPi Instances = processor/HT cores (max is 16 instances)", logically it's not clear that you may not disable cores or HT, it just says bench instances must be equal to core/HT number. Personally made this mistake, adding clarification probably will save time/effort for others
Easy way to ban lga 1200 without banning 1200 lmao
-
16 minutes ago, NATA 58 said:
We try to put tests where enthusiasts can compete.
There is no bench where enthusiast is more or less competitive, except maybe pcm10 because ssds don't need ln2.
- 1
-
31 minutes ago, unityofsaints said:
@Leeghoofdthe Time Spy stage is currently a mess, in the design thread it was mentioned that Titans and all RTX cards would be banned but as configured, some RTX Super cards have been missed and some older Titans are allowed, e.g. RTX 2060 Super and Titan Xp. Can you close the gaps please?
You just mad cause US isn't doing bad lmao
- 1
-
I say take all the legacy globals, if a bench doesn't have support how can it be fixed and kept secure.
-
31 minutes ago, FUGGER said:
Yos, I am not running ramdisks. Strong disks and cache go so far as you know, it takes a well rounded machine and too much time to re run and tweak.
Exactly, am in awe of the raid arrays you pull out. Ram cache helps but still need fast disk, ram drive no need.
-
2.5 is a preset but it uses like 16gb of ram, hence my statement on the paradox. 1b is too fast on big bois, 2.5b uses too much ram on smaller cpus.
-
14 minutes ago, FireKillerGR said:
3D sucks for being more demanding than most if not all 2D benchmarks?
Was mostly tongue in cheek, personally I hate doing 3d cause I'm bad at effi, but I think for those that can enjoy it that's awesome.
As for legacy benches, I think no matter what 3dm03 should not have globals, it's extra buggy and I have my own reasons for hating it as far as pts are concerned. Those that know know, those that don't, well I cant elaborate lol
- 1
-
5 minutes ago, Rauf said:
One thing to consider is that sli and crossfire support is all but dead. As far as I know amd does not support it at all since rx5000. And nvidia has dropped support but it still works for a few newer benchmarks, however only in 2 card config.
That means we have one category of globals per benchmark (1xgpu). For a few combinations of benchmarks and vgas we have two categories (1x and 2x gpu). For 2d benchmarks we have 8 categories per benchmark (4, 6, 8, 10, 12, 14, 16, 18 cores). And that's just counting the most commonly used. The variety and total amount of available globals are way, way more for 2d. If these is a desire to "compensate" this, maybe there should be more 3d benchmarks that give globals compared to 2d?
Sounds right, 3d sucks, frankly I think 2d should have MORE globals
- 1
-
30 minutes ago, speed.fastest said:
I think global need to move to R15 because R15 can run on x32 cpu too if im not mistaken.
Does r20 not scale?
-
Globals for 3dm01? I don't like it but I do get it and suspose we should lol
Catzilla globals idk, bench is a bug fest and devs are not actively supporting it.
And can we not drop night raid? We have almost all the 3dmarks, wouldn't make sense to drop one that has active support
- 1
-
4 hours ago, Rauf said:
No idea how ycruncher works but it is AVX right? Needs lots of mem also?
It even uses avx512, 7800x is relevant again! But yes it loves mem bandwidth, on 3175x core was practically irrelevant. 1b needs about 5gb to itself, so probably a 6-8gb maxmem. Imho it's kinda tough cause on big bois it takes under 20s, but if we went up to 2.5b it would be way too much ram for smaller core counts.
If you want to compete for 10b hw pts you need as much ram as possible basically, bench uses over 40gb on its own.
- 1
-
4 hours ago, Rauf said:
Pcmark? Never tried it, are they "safe" or will they scale immensively with super raid configs etc?
R15 will be missed, but I guess it is a little fast on current high end platforms.
No idea how ycruncher works but it is AVX right? Needs lots of mem also?
I would like to see some 3dmark cpu test added to globals. Sad that time spy does not really scale with core anymore cause it's a good mem intensive benchmark. 3d11 cpu is also good cpu/mem benchmark, maybe an option? Have been tried in competitions and works well. Time Spy Extreme cpu test could also be an option.
Pcm always scales very much with strong storage, even with ram cache. Just look at fugger's work lately lol
- 1
-
> hwbot prime 30m and system stability test removed
Nooooooo my pts! Although in all seriousness good riddance
Now pyprime was recently added and the dev trying to get it up to snuff, I suggest we leave it in beta as long as the dev keeps working on it. We need more singe thread benches
- 1
-
5 hours ago, Leeghoofd said:
Whats the difference for a ram disk and ramcache for the PCM program? you are not actually testing true storage like HDDs or SSDs...
So with ramdisk you store the whole program in a software ram drive and it will never hit ssd, you also cannot install an os to it which I think is why hardware ram drives like iram or acard are allowed. Ram cache is more of tiered storage, even with a large persistent cache it has to hit the ssd eventually. Ram cache is common software tweak irl because it let's you get improved performance for certain smaller files you'd hit often without the overhead of needing insane amounts of ram to store everything. Even amd uses this with their storeMI tiered storage for Am4. If this were an ssd bench I'd say ban ramcache, but it's a total system bench so it's about how fast your total system works together and how efficient the os is for "productivity and daily tasks" so I think ramcache fits that.
End paragraph
-
2 hours ago, Jumper118 said:
That is strange the i7's seem to do upto 6ghz of the same gen.
Issue is probably me, 1650v2 was suffering too lmao
-
1 hour ago, Leeghoofd said:
Don't waste your time on 05 anymore... It has been ripped apart by all the software tweaks/hacks. This one will become completely bointless. No use to keep this as a guy on air with right software can outperform an ln2ed setup. Users need to learn that benchmark software needs to be run out of the box. Also spread the word: the users that have a cheatengine icon on their desktop will get hammered as well. Tired of this "trying to break a benchmark and finally spoil it for the others attitude".
But is ram cache allowed? Pcm is supposed to measure the overall performance of a PC for actual tasks, or at least to the extent of when it launched, and ram cache is a daily viable tweak that can help actual storage work better. Just asking because it's allowed for other pcmarks but it's ambiguous for this one
-
Is ram cache allowed like in pcm 07? It's called out as distinct from ramdisk there but not mentioned here
-
STRONG did you try enabling large memory pages?
Forks - Phenom II X4 960T BE @ 6180MHz - 2.1 fps HWBOT x265 Benchmark - 4k
in Result Discussions
Posted
I'm proud of you son