yosarianilives
-
Posts
2268 -
Joined
-
Last visited
-
Days Won
9
Content Type
Profiles
Forums
Events
Blogs
Posts posted by yosarianilives
-
-
2 hours ago, Leeghoofd said:
and for that reason it mentions sandy.ivy and haswell, no Broadwell for that stage, I'll exclude the 1151(v2) and 2011-3 from DDR3
Can we drop haswell then? Cause now it's just gonna a haswell comp (4770k+4770r+4980hq), I feel like if we're trying to avoid people complaining about needing "rare" 5775c then we should make sure that it's not about the even "rarer" dual pcb mobile chips on s1150.
-
Also secondary note, are we allowing broadwell igp? I could see arguments both directions, personally since we're already allowing s1150 haswell cpus we should allow broadwell, however it would basically just be a broadwell competition then.
-
3 minutes ago, Noxinite said:
You might be able to, but only for the iGPU stage (are there Xeons with iGPUs?) or the Dogpile stage.
Yes now that r15 has been narrowed down, glad that the winning combo isn't 9900k+8700k/5960x+6950x
-
1
-
-
1 hour ago, Leeghoofd said:
Xeons are allowed up to 6 cores, I added a lot of limitations to the compo itself
So this means that we can run sandy/ivy xeons as well as haswell/broadwell xeons (shhhh). Also similar vein, if I somehow get a 9900k or 5960x/6950x working with ddr3 those are allowed in the less specific stages?
-
1 hour ago, Noxinite said:
Thanks for the clarification.
Edit: Oh, also what are we doing about adaptors?
I assume it should be same as things like s775 spanning ddr1-3, as long as the cpu's socket doesn't transgress the limitations (s771 in s775 is still a s771 cpu) and the memory type is correct it will be allowed. I don't think I've ever seen an adapter cpu count as the motherboard socket instead of the cpu socket.
-
42 minutes ago, unityofsaints said:
Not a single SP32M stage? System memory restricted to DDR2 on a GPUpi for GPU stage? Dogpile stages? A 3D stage restricted to 7 really expensive cards, but not the 8th compatible card because that's too expensive somehow? 6 repeat stages from last year?
- DDR1 3DM01 AMD only
- DDR2 CB11.5
- DDR3 3DM11 AMD only
- DDR4 Geekbench dualcore
- DDR4 Firestrike APU
- DDR4 Aquamark AMD only
What is this, a deliberate attempt to kill off comps once and for all?
You'd think that a 27-stage comp would have something for everyone but this TC has managed to circle the square, against all odds
It is kinda sad to see no 32m but the repeat stages are repeat cause they're too good to not revisit
-
1
-
You missed the opportunity to use the glorious phrase "piccy with a sticky"
-
1
-
-
16 minutes ago, ground1556 said:
For memory frequency stages, do the used CPU or used memory have to be unique for submissions?
Believe he said unique kits
-
For the port royal stage will rtx titan be allowed? Or will max be rtx 2080ti? Since it's single stage and single score I figure it would be like the one sli stage of last year.
-
I'm gonna have to avoid the temptation of trying to buy 1x of an entire socket of cpus to drop at least a stock or ss sub for every chip
-
31 minutes ago, mickulty said:
Speaking of loophole hardware, is it intended that "3dmark Firestrike integrated Radeon graphics (2 scores) -> max 25 points" would include (or could be interpreted as including) Vega M?
Good question and hopefully the events of last year would convince all that it should be clarified to amd cpu+igp
-
1 hour ago, TAGG said:
That's not what i meant
I know there's 2 deneb chips lost in there, but then again that's probbably why the limit is AM2 (non plus), so theres no ambigious CPUs
You know finding loophole hardware is my favorite part of these comps, mainly because unlike actual benching I can do research while bored at work ?
-
3 minutes ago, Digg_de said:
This Year its GB4.
Much different very bench
-
I like the changes that were made, they seem to adequately answer the suggestions. Only thing I currently wonder about is lga 1151 Ddr3, although I suspect it'll be answered in similar way to last year by stages having sockets specified. Tbh not even sure I don't want to see lake Ddr3 since it would take a lot of work to make it really good
-
18 minutes ago, TAGG said:
We had those hw series last year, what do you think about geforce 7 series and AM2+ ie. original phenoms? Don't think they've been TC for a while
I love an am2+ stage, especially cause there is a few loophole cpus that aren't even hard to get (will leave that up for research, but look closely at the phenom list on wiki)
-
1 minute ago, mickulty said:
So... we need a 1366 stage?
I do have one question/comment, anyone know if Black Hole is competition ready? I've had trouble getting consistent/competitive results but maybe I just suck, I'm happy to accept that (-:
I've heard blackhole is rng so perhaps not?
-
3 minutes ago, havli said:
Hint for the X265 4K Dual Core. Perhaps the locked Skylake i3 / Pentium / etc. with BCLK overclocking and disabled AVX might be faster than almost default with AVX2. But who knows...
Current scores say no, pretty sure we researched this for previous TC with basically exactly the same stage and it ended up being winning combo was 7350k+7350+another i3, all with 102.95/8 bclk for the locked chips
-
Comments to right
DDR3: (no server cpus allowed)
- 3DM05 LGA775 and GeForce series (3 scores) - can we allow 771 xeons as well?
- 3DM Ice Storm - Intel IGPU (Sandy/Ivy and Haswell) (3 scores) - big thanks! If we're allowing haswell then maybe broadwell too? If not it makes sense too tbh
-
3DM11 Entry - AMD only (3 scores) again is ati allowed or only cards after amd buys ati?
- Black Hole Benchmark (3 scores)
- GPUPI v3.3 for AM3: 2 - 4 and 6 core subs (3 scores)
-
CB R15 (3 scores)
- DogPILE X265 1080P (999 scores) - can we allow intel server cpus for pre sandy?
DDR4: (no server cpus allowed)
- HWBOT X265 4K - Dual core (3 scores) - there's only one cpu that's gonna have hardcore OC for this stage, the rest will be hardcore effi
- Geekbench 4 Multi with Benchmate 4 - 6 - 8 score (3 scores)
- CB R20 with Benchmate (3 scores) no TR and LGA 3647 - so basically just x299? Makes way more sense to allow tr4 and allow max of 16 or 18 cores imho
- Aquamark3 , AMD only (3 scores)
- 3dmark Port Royal (1 score) -> max 25 points
- 3dmark Firestrike integrated Radeon graphics (2 scores) -> max 25 points
- Superposition 1080P Extreme - no 2000 series (3 scores)
-
Comments to the right of stages
DDR1: (server cpus allowed)
-
Unigene Basic Heaven: 1 core CPU (3 scores) - Should see some interesting combos here, l3014+asrock 775 ddr1 board+rtx 2080ti
- 3DM2001: AMD only (3 scores) - So assume this means amd cpus +ati/AMD gpu? Or does it require gpu made after amd acquires ati?
-
3DM99: Socket 478 & AGP GPU only (3 scores) - Glad to see an agp stage, this is always a fun challenge
-
CB 2003 (3 scores) - I think we know the ideal mobo and cpus for this stage
- SuperPI 1M Socket A (3 scores)
-
Reference frequency (3 scores) - glad to see there will be crazy p4 scores
- DogPILE WinRAR (999 scores) - will this stage allow multi sub from same person? I feel like it would fit cause it's not like a single person can do this stage either way like you could for some limited sub stages
DDR2: (server cpus allowed)
- Geekbench3 Single core: AM2 platform only (3 scores) - as a fan of am2 thanks bunch!
- Wprime32 - Odd cores only (unlocking allowed, disabling not) (3 scores) - will this specify that you need each odd core count? (1,3,5) If not it should else it be a different flavor of the no thuban cine stage last year
-
Cinebench R11.5: (3 scores) - to see if yorkfield or thuban wins, but I suspect thuban. Glad to see this stage
- 3DM03 for Geforce 6 series: (3 scores)
- 3DM Ice Storm : VALID verification link required (not unrecognised CPU or other stuff, only VALID counts) (3 scores) - ditto to r11.5
-
GPUPI 1B (3 scores) - ditto
- DogPILE 3DM11 PHYSX (999 scores) - ditto to other dogpile stage
-
Unigene Basic Heaven: 1 core CPU (3 scores) - Should see some interesting combos here, l3014+asrock 775 ddr1 board+rtx 2080ti
-
18 hours ago, Noxinite said:
Weeeell, not always. ?
Exceptions prove the rule
-
5 hours ago, Noxinite said:
No, 430 is garbage for Conroe. But you don't need it for this chip as high multi.
Ah, so more about binning core than fsb. Whereas 9x chips it's all fsb and you can often take gold on ss.
-
@Leeghoofd Any word on when we'll hear stages for this comp? Understood that it was going to start in August, possibly before the PA bench meet, and now we're approaching the last week of July.
-
3 minutes ago, Matsglobetrotter said:
Geek bench starts as 32 bit but I select 64 bit inside. The problem i had was anything listed as 32 bit in benchmate would not start. now they do. Thus I believe the service will capture it right.
I believe my issue is gone as i cant replicate it anymore even after now 4 reboots. Meanwhile if i run Geekbench4 in 32 bit more i loose about 600 or so points on single core performance.
The same tweak to get geek 4 to run in 64 bit should work
-
I don't know a lot about 65nm, is 430 fsb particularly high? I know it does less than 45nm
The official HWBOT TEAM CUP 2019 thread
in HWBOT Competitions
Posted
So that means skylake and kabylake are still allowed in the dogpile stage? Only CFL and hwe/bwe are disallowed?