yosarianilives

If you don't have a key then you need internet connection

I need to compete on this

Ygpm

There's always legacy. These kind of cards will always be cheap when this old

Noticed the round 1, we'll see how this comp goes but I like the idea of benches staying very similar but doing more rounds with some of the other suggestions from this thread.

Let's start tomorrow since everyone seems to be ready. End with still be end of March

15th sounded like a round number, but we can start before the weekend so people get an extra weekend.

Yes

How about, and here me out on this cause it might sound crazy. But how about we make pts be based on your ranking vs other people, and use a fixed points curve? Sounds pretty crazy so idk if the community will like it

For anyone wondering details, plan is to start comp Feb 15th and run until end of March. Cpu limit will be 5005 mhz with no other restrictions on cpu, tests will be 3dm01, 3dm03, and 3dm vantage gpu score. We will allow all variants of GT218 core so to my knowledge the allowed cards and cores in DB would be as follows: geforce 210 ddr2, 64 bit, gt218 geforce 210 ddr3 64 bit ddr3 8400 gs ddr3 gt218 geforce 310 NVS 300 quadro fx 380 LP Cores allowed: GT218 GT218' GT218 GL Let me know if there's any thoughts etc

Going to be based around gt218 core (gt 210 card as well as a few cheap quadros etc), closing the poll for bench selection as I think we've gotten all the votes

If I can get globals by running stupid server cpus that have 15 cores each in multi socket I don't see any reason not to give multi gpu globals. The only exception I can think of is geek 4 compute as it doesn't use sli/xfire and it does not support multi gpu at all

Bad effi tbh, didn't even test different Numa configs. Maybe I'll do more testing if the 8895 works

300k is possible, not sure if this gen

What would it take for 25 in 4 core? New arch or ln2?

Time to get the popcorn out, we about to see a battle

People complained that there were too many cpu pts and not enough gpu points. Of course maybe those complaining have cards lol. At least 3/4 way shouldn't be expensive if the market ever fixes itself

I like 3dm03, my only opposition is that it's easy to accidentally bug if you're on edge of stability but luckily it's easy to catch in the screenshot if it crashed in a subtest. Forks is gonna have to force me to mod a card lol

Yeah max multi should be a bit lower, but not all the way down to base multi

Based on discussion and my own opinion I'm planning cheapaz 2022 around the Gt218 core (Gt 210 and several other cards) so benches will be legacy, gonna shoot for 3x stages with some reasonable cpu limitations tbd.

Thanks, still have a lot of testing to do. Triple socket is cursed, windows want to make uneven processor groups despite there being three Numa nodes it enable ht.

A new challenger approaches

Wondered how long until someone gets this. Now the effeciency race is on, wonder if there's much variation on imc or just how long until a "good" ddr5 ic comes out

I think we would do as previous where we just specify a core, so say any gt218 and if people find a card with that core, even gt 405, that somehow has some super awesome memory bus they can use that. I think it won't be too hard to find the optimal memory config cards whatever that ends up being. And all gt218 is the same core config so no issues there. I'll give it a day or so then put up a poll for benches and for card. I think the two contenders are gt218 core (gt 210) and oland core. Benches will be legacy, especially if we go with tesla arch

No 12300 yet?