Minotaurtoo

In Minotaurtoo.  I hope I don't forget lol...

9 hours ago, BWG said:

Stats were broken last month. Prizes could carry over to this month if that's ok with everyone?

 

Damric asked me to run fat this month and next. Axipher still has a stats issue last I saw. I'll bug the powers that be more to see what they would like to do for August, and then put the thread up.

fine by me, I only was hoping to see what I actually averaged over the 2 days lol... (I forgot to look at my total before I started)

I'm still at the edge of my seat waiting to see what my 48hr points tally is : )   Those new drivers really made a difference

I'm a bit late here, but having lived in that area and knowing how those streets are... I'm amazed there aren't more accidents like this... I lived on the north side of town on Merrimon and man I can tell you getting around could be a nightmare at times back in the late 90's... I usually stuck to side streets when possible...

well as the finish line is approaching I've started getting some nice work units and my ppd estimations have been holding at 5mil +, way better than with the old drivers.  Surprisingly the cooler than normal temps outside held long enough for me to make it the whole time. 

well so far I've discovered that my undervolt on the gpu isn't 100% stable, but otherwise so far so good.... here's looking at the finish line.  This winter I'll be back to folding as normal, for the heat.

IN as Minotaurtoo  Temps outside cooled down a bit so I figured I'd join in, may have to tone it down a bit or even turn off FAH for a bit if it gets too hot... as such if you want to exclude me from prizes, by all means do... I really don't need them anyway and would rather see someone who needs them get it.

I asked it for a city made out of computer parts.

wonder if it fixed an issue I had where I had to disable smart access memory or I'd have a driver crash when folding.... doing a short test now... so far PPD estimates are way up... I may join in the comp just to see.

1 hour ago, ENTERPRISE said:

It is indeed back better than ever. Lots of diagnostics and backend improvements. The downtime is always a pain but this was necessary to fix a few bits and pieces reliably. Thanks all for your patience.

Thank you for all your hard work keeping this site going.  Love it!  Way better than the "other" site that would take years to do nothing about a known problem.

when given the right prompts this thing's pretty good... here's Machu Picchu in the clouds

@bonami2 the thing seems to have a unique way of interpreting what you want lol... I love it though.

Flying cattle in space

Bull at sunset in thunderstorm

Nightmare sunset

large waterfall into lake surrounded by fir trees

10 hours ago, Mr. Fox said:

That is so true. Some (many) are just too lazy and others don't even know how to read. They just look at pictures.

yep and even within companies comparisons are only good for exact same hardware... here is a fine example... the 9590 at first glance looks better than the 1700, but looking across at the sub scores shows that's not the case.  The problem as I see it, is that userbenchmark calls it a "gaming score" in the processor area up top... now we know that a 9590 isn't better at gaming, or anything for that matter, than a 1700.   full disclosure, both were overclocked mildly.  This is the only benchmark that I saw the 9590 show a "lead" in.... but as has been stated, it just shows the full story is to the right of that "bench" score percentage.  If I had to guess, I'd say it was due to the severely increased latency that ryzen had with memory.

Links:

1700: https://www.userbenchmark.com/UserRun/6449742

9590: https://www.userbenchmark.com/UserRun/4758334

On 15/07/2022 at 15:36, Storm-Chaser said:

You nailed it Mr. Fox. I just didn't want to come out and say it LOL.

 

And I totally agree, there is no bias against AMD with this benchmark. Moreover, they have been very kind to AMD in the past, which further demonstrates there is no real bias here. 

 

That being said I've had this benchmark competition shot down in flames MANY times because lots of people think it's way bias or a total joke. And I'm talking about a solid majority here. Or, they don't understand that core count isn't the "BE ALL" of CPU performance, so they wonder why an 8 core intel chip ranks higher than some AMD 12 core chip, and they just assume it's total BS because of that.

 

So I've found the only way to run a successful competition with this benchmark, because the hate for it is so strong, is to bash it and confirm their flawed suppositions about it's bias in the very first post. Basically, I have to destroy the credibility of userbenchmark.com before anyone will actually be willing to jump on board and participate. If I don't do that people try to destroy my credibility and bomb the thread with laughing emojis and stuff like that. 

 

And it is a shame, because like you said there is a good bit of value in measuring your system in this way. So, is userbenchmark perfect? Of course not, but what benchmark is? In my opinion it's simply another tool in the toolbox for measuring system performance (in this case against like hardware). 

I love userbench for quick dirty benchmarks when comparing like hardware... but beyond that, I look elsewhere for intel-amd or amd-Nvidia comparisons... I don't think it's as bad as people say, a benchmark is just a benchmark and I don't think anything nefarious is going on.  However, the person in charge of reviews is biased and often on an ati-amd rant...

Just read these reviews to see what I mean:

https://cpu.userbenchmark.com/Intel-Core-i5-11600K/Rating/4113

https://cpu.userbenchmark.com/AMD-Ryzen-5-5600X/Rating/4084

Now don't get me wrong, that's totally their prerogative, but it does mean I will take them far less seriously than I otherwise would.

On the core count argument... I find that funny because many of intel's own high core count chips fell way down the list compared to their own 6 core chips in the past... right now on their current generation, the stack up in Intel is pretty much following the core count, but I think that's due to Intel's better binning process on their latest chips.

For those who think that there are no real life examples of amd getting their butts handed to them, just look at FAH... while at gaming I find the 6900xt to be pretty much on par or slightly behind the 3090 in most case with raytracing off, in folding... no... it falls way behind.  For me, the 6900xt made sense... just happens that my favorite games do better with it and at the time it was way cheaper....I actually got it cheaper than a 6800xt at the time don't know what was up with that.

ok, it's old, but here's one of my best overall tunes: https://www.userbenchmark.com/UserRun/43615155

8 hours ago, ENTERPRISE said:

 

@MinotaurtooApologies, I forgot to get back to you. The donor badge is for those who are donating toward the running costs of the site  Members with the badge will have donated by selecting a donation option HERE.

Done!  It's not much, but I'm glad to help out.  Love this site much better than other tech sites that have been on before.

cool looking badges ... I participated in this too and as of now do not see a badge... also... what do you have to do to get the donor badge?

It's not a pc part, but cost as much as many pc parts.  My wife surprised me with an early Father's day gift...then asked for food lol... I spent most of the day yesterday seasoning it before finally getting to grill burgers last night.  I don't use propane much, but I still hooked up a tank.  I usually grill with charcoal or wood...

For those who wonder, seasoning a grill like this is basically coating it with vegetable oil and getting it warm for hours then hot and done... the bottle seen in the pic is the vegetable oil.

2 hours ago, bonami2 said:

Seem like they have a hard time making those work.

 

I need to do more testing but am lazy for now gear 2 seem to run at stock xmp. DDR4 was out for so long and they never figured how to make 4 stick stable at high clock 

I managed to get a good boot on 3733 and a not so stable boot on 4000 so yeah, must be a hole in what this board can handle... but strangely I have to decouple the memory controller from the IF to do it for either with leads to lower performance... so far 3600 with tight timings is king... I think it's just a board limitation, might try swapping to the other memory slots to see if they work better... board traces may be crossing over or leaking.  As for your 4 stick issue, the most I was ever able to get out of 4 sticks stable was 3200 in gear 1... never tried gear 2 to see if that would help.

well drat... that's ram for you.  I found out that my board apparently has a memory frequency hole at 3800mhz.  I looked at the memory support page for it and the supported memory goes from 3733 to 3866. not one kit of 3800 listed.

8 minutes ago, J7SC_Orion said:

 

What 'native' CL rating is your Samsung B-die 3600 kit -  CL16 ? CL14 ?

 

Anyway, you might try 1.5v (ok for B-die if there's enough airflow) and 3800 with 14-14-14, or 14-15-14.

My kit is a cas 14 kit with stock timings of 14,14,14,14,34....  I actually managed to get the dram calculator to work finally... after a re-download and using Thaiphoon to get my xmp to import into it.  I still can't get 3800 working though even with the dram calculator finally giving reasonable numbers... I was thinking my Fclock couldn't hold up to the 1900mhz needed, but then I was looking at supported RAM for my board and there are no 3800 kits listed... just skips from 3733 to 4000 so maybe it's just my board has a "hole" in frequencies it can handle on the memory traces... IDK. but with the tight timings that I have now, I'm feeling pretty good about just keeping it at 3600 14,13,13,13,26.  

glad got that figured out.  I'm still playing around with mine to see if I can get 3800mhz to boot... it'll post, but it's not stable, even in bios.  I haven't tried the "safe" timings yet from the dram calculator, but the way it acts... I kinda think it's the infinity fabric losing stability... IDK, new to this level or ram speed.

27 minutes ago, Avacado said:

Glad you figured this out. I knew your B-Die could hit 3600 easily. Can probably do more if you ever wanted. 

There's actually a bit of a story here... It's not the same B-die ram.  I think I told you about how I was planning on practically giving my old 5700xt away by selling it dirt cheap... well... the young man couldn't buy it even for that cheap because his Jeep had just burnt up completely... so I just gave it to him outright.  We were talking about me trying to get 3600mhz on my ram but I just couldn't get it to hit it for nothing and he said "you know what, I have a spare 3600 b-die kit"  He had a set of 3600 cas 14 2x16gb  He sent them my way.  They run at tighter timings @3600 than my old kit did @3200... So yeah, kindness sometimes comes round.  Speaking of that, I've already found someone who wants my old 3200 x4 kit... think I'll just let them have it.  

Ok, problem solved.  I deleted my old Dram calculator and re-downloaded first just as a precaution (thinking it might have saved some settings)  Next I downloaded the freeware version of Thaiphoon and exported the xmp profile for my ram... next ran Dram calculator again and imported the xmp profile... hit calculate fast and voila, working timings.  I have included a screenshot in case anyone wants to try them.  Going from 3200 to 3600 gave a 5-10% improvement in many places I didn't even expect, like cinebench...