yosarianilives

I mean you could consider hwboints a competition in itself, and so including benches that are questionably about overclocking into hwboints is unfair for those that would like to just OC and get hwboints. Stripping an OS is one thing, doing everything but change the .ini is another.

In my experience most software tweaks in legacy benches would be considered cheats if people chose to look into what they actually do. Great example being monitor pull in AM3 that was accepted as a tweak for years until it was found to just be a way to bug the bench or how LOD/Tess discard is allowed on old benches because it's undectible and not because it's considered "not a cheat". There's a reason it's not allowed on newer benches that can detect it. I think far too many tweaks are considered legit because people would rather be willfully ignorant to the possibility that their good score is due to a bug and not that they're a s00p3r l33t tweakzor

I don't like to compete in them anymore because at this point being competitive is more about who can bug them harder and not actual OC.

Another year and hopefully another Cheapaz Chips. Unfortuneatly AMD and Nvidia have been neglecting the low end the past year so we don't have a lot of options. So far it looks like potentially we could do RX 550 or 560, GT 1030 DDR4 edition, revisit GT 1030 (PLZ no), or do something with APUs (there's no good mobos for am4 APUs so still would benefit from mods). I've created a poll, if anyone has any other ideas for potential hardware comment below.

Not sure on the cost but I know of at least a few people who would consider contributing towards Mat creating a wrapper depending on the cost.

Well it looks like I owe you all an apology for being an idiot. I was doing some testing on my 7700k for hd graphics 630 to try to get some vindication and was getting really confused why my apex could only do about 1450 mhz on the igpu core when my soc was running "1926" mhz just fine. So I swapped back to the POS z270 soc. Boot at 1500 mhz multiplier just fine and I decide to double check clocks with gpu-z. To my surprise it read 1150! So as it turns out igpu multiplier is completely broken on the z270 soc, add that to the list of things that don't work right next to 1t cr. So I did the math for the bclk I was running with the multiplier I was actually running and it comes out to just about the same mem mhz that I was running on the apex. So it seems that on ss 1500 mhz really is the igpu core limit more or less. So I'd like to apologize for suggesting that you guys might've been doing something wrong, when it was really that I was an idiot the whole time. (as I suspect many of you already knew) Also after my testing yesterday I can say that it seems that the subtimings priorities are not necessarily exactly the same as on ddr3 platforms as I didn't see the same kind of scaling I'd expect. So while mem is definitely important I can't say I have a clue what you need to do other than have fast/low latency ram. Now if this was ddr3 based intel graphics I could give a fair few more pointers Hopefully I can learn a bit more about this part of my niche as I work towards becoming the intel igpu king (intel gma will also be interesting, I have about 15 boards and plans for dice lined up)

This now links to the microsoft store, on this basis I'd say that kills the bench for hwbot until they can fix it. Especially since it's not any more secure to .ini "tweaks" than previous versions of cine are.

If we're lucky it'll lose all points. That pile of garbage that intel calls a bench should've died a long time ago.

AFAIK you usually need windows management instrumentation enabled for it to work.

Ladies, enough of whatever catfight/feud this is. Can we stay on topic of how leagues should be seperated?

There are additional requirements to enter the comp, must also read rules on competition page

The point of this bench is to get cpuz to read a number, has nothing to do with comparing to other cpus in any practical sense. If you want to do that look at a bench like x265

You have completely misunderstood the point of this bench.

A cpuz validation doesn't even run any stress, it's literally a test of what clocks cpuz reads out so it is in fact single threaded

They no longer add devices, only processors, this breaks the mobile hwbot prime submission however that software is deprecated and should be considered EOL with no support.

There are many "singlethreaded" benches out there like superpi 32m which people tend to disable cores for. I don't see any reason to do away with that.

Lol, this times it seems it doesn't matter. As long as you don't get a ddr2 model or something and there's enough vram for the bench (all models should have enough) it should just be about mods (I'm sure there's an IC that's best but I don't know what it is)

No longer true now that avx2 is supported, most cpus haswell and newer it impacts. As for gpu tests it shouldn't matter but :shrug:

Yes, HD 5450 or any other Cedar core

There are still many timings that are hidden on both intel and AMD, so perhaps you may still be able to pick up a little extra effi by prioritizing bclk over multi

I think we should take cooling out of the equation plain and simple, should just be rookie (first 6 months) and "enthusiast" (>6 months). Verifying cooling is too hard to fake, and given the recent push to make all benches cheat proof with wrappers instead of just a screenshot I think separating leagues by cooling and verifying with a rig pic is backwards thinking.

I'd like to move towards that. It must be valid on 3dm site (only excused failure to validate being unapproved drivers), I'd like to see this be more about the hard mods and less about driver mods Speaking of tweaks, nobody knows something super easy like monitor pull in 3dm03 that would make one of these benches a shitshow right?

Unless peoplre throw a big fuss in the next month I think we're going to go with all cedar cores and the top 4 benches.

I'd like to go with Cedar core

Since it's exclusively the graphics tests core count doesn't matter for the selected tests.