Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

bmichaels556

Member
  • Content Count

    293
  • Joined

  • Last visited

Everything posted by bmichaels556

  1. To be honest, I'm a total layman with this stuff under the hood, but from the little I know, ARM processors are absolutely insane... You'd basically have what, mid to high-end x86 desktop performance in a 15W TDP laptop? Far as I know, smartphones are not far from a decent desktop just processing-wise. I do wonder how well they scale up to using more power, higher frequencies etc. Like, would a 100W ARM processor with an insane amount of processing power be feasible, or is there any kind of limitation? I guess you could add a metric shit-ton of cores, but surely you're going to hit a clock speed limit just due to the silicon? I'm pretty interested to see where this goes though. If they're going to do it, then others will have to make moves too. What would it look like for everything on the desktop we're currently using? Could an ARM port of Windows use emulation to run pretty much any x86 programs?
  2. I ended up pulling the trigger on a Thinkpad T440s (w/ i7-4600u and 8GB RAM) I found for a good price and put in a Crucial BX500 2.5" SSD (240 GB). IIRC, this laptop does actually have an M.2 slot (I think) but I had the BX500 on hand for another project I didn't bother with. Problem is, it feels quite slow. All the benchmarks I'm looking up are showing significantly better numbers than my drive in both CrystalDiskMark and AS SSD Benchmark. Actually, the benchmarks I had taken when I first got the drive (basically empty, after new Windows 10 install), The numbers seemed to be where they were supposed to. For AS SSD Benchmark - Now at about 56% full (installed a number of games to dink around with for no reason), 4K 64Thrd is down to like 1/3 of where it was, and write latency has tripled! Is this normal? I know a fuller drive doesn't perform as well, but that seems like a massive performance disparity. For CrystalDiskMark, things actually look fairly good, mostly where I'm seeing the numbers from what I'm looking up. But I dunno, something about it just seems kind of sluggish and I don't know why that is. Even my old Dell i3-4010u with some an ADATA SU630 felt as fast or faster than this thing. Chrome also acts weird and I get weird video freezes and audio glitches, would could totally be a driver thing. I mean overall I like this laptop. It COULD be the case that because I have all the Intel power saving stuff enabled in the power options but I've tried all of these things both plugged in and unplugged. What are your guys' thoughts on this? My desktop is a dual X5675 w/ 24GB RAM and an RX480 and this thing generally feels blazing fast. I figured despite only 2 cores 4 threads, the T440s would be pretty damn snappy since it's what, three generations later? I wonder. Should I pull the trigger on a cheap NVMe drive? IIRC, this laptop only has two lanes for it (maybe a Haswell thing?) but I wonder if it's worth going that route.
  3. Wow, those numbers are impressive! Really efficient, that's crazy. I really need to pick up a 5700XT once I update my setup. Are you able to check your sustained clock speeds in Radeon Settings or whatever other program?
  4. Hey LTT gang. I'm looking to get a good amount of info on the PPD you're getting with assorted hardware. CPU, GPU, settings on them and so on. My experience with my RX 480 8GB (got rid of many others I had, so no chance of testing with multiple cards) is that I'm seeing about 425,000 PPD stock (it's a Sapphire Nitro). After lowering power limit 40%, I was doing about 365,000. After just a bit of tweaking (raising memory clock to about 1900 or more while leaving the -40% power limit) is giving me an amazing 390,000+ PPD with only about 82 watts of power usage. I'll probably fiddle with things a bit more to see if I can get even more points per watt but I'm already impressed with this thing. In fact, I did even better. Small undervolts along with the aforementioned settings are giving me right on the edge of 400,000. CPU results have been interesting. I'm using dual Xeon X5675's (12 cores, 24 threads total) and I'm more or less seeing the same numbers with either 12, or all 24 threads folding. I wonder if maybe it's cache-related or something like that? Similar results on my laptop (i7-4600U). One thread is giving me well over 5,000 PPD. Two is giving well over 8,000. Three and four shows absolutely no increase. If anything, all four might be showing a decrease. But that's not too crazy considering it's a 15W TDP processor. I'm really curious what numbers others are seeing and would love to collect a lot of that info for the last two gens of both AMD and Nvidia cards (and your settings and PPD with different settings you've tried), along with any CPU's you might be running, how many cores/threads and so on. I'm mostly just trying to collect this information for my own curiosity more than anything. Like, literally any hardware you've tried to fold on would be awesome to know about! Older stuff included. I just happen to predict that the older stuff, maybe R9 200 series and prior HD 7000 series are probably not as attractive, maybe with the exception of the 290 and 390. But I'm rambling. Thanks guys!
  5. Hey gang. Anyone have a big list of PPD for different cards? Hopefully from the last gen or two of hardware from both AMD and Nvidia? But any of that info would help me out, even if it's a little incomplete.
  6. Seeing awesome answers here! I think it comes down to this: With Ryzen in particular, if you're using something like 2133 RAM, you're gimping your system for quite literally no reason. I believe after ~3200MHz, the performance increase is close to zero and cost is too high. But for a new system, it seems that 3200 is a nice place to just "set and forget" for best overall performance increase, at least from the data I've seen! I think Tech Deals did a really good comparison from what I remember.
  7. ... I can't even get my 1 GPU and two CPU's to receive work and fold and send and be credited correctly... "How dare you!?"
  8. Ugh... Over the past like 5-6 days of folding, I've only managed to actually end up with about 1.5 days of credit (looking at PPD). Is there a big delay from when work is submitted to when it's counted in the stats? Also, could it be other factors aside from too few work units, like not sending correctly or something like that?
  9. I do wonder how the HBM cards fare here. I remember back in the mining days where you could chop off like 50% of power and lose only like 10% of already huge hashrate. At least for Ethereum as I recall. In fact, all this really does remind me of my mining days.
  10. Random thing I noticed, but at face value it seemed like my experience with an RX480 8GB was that core clock seemed to matter quite a bit. Stock settings gave me a little over 400,000 PPD. But I've been doing -40% power limit and only lost about 15-20% of PPD, so that was cool. So about 365,000 average, but sometimes dipped as low as 350,000 PPD. But then I started messing with memory speed while maintaining the -40%. I went up to 1950 Mhz on memory, which automatically brought core clock down from about 1175 Mhz to about 1075-1100 Mhz. I suppose because the memory doesn't use much power overall compared to the core? But now my PPD is only just under stock settings. About 390,000 PPD average. So memory speed seems to matter quite a bit here, and you can REALLY dial down power limit and just keep memory really high from what I can tell for extreme power savings but only a small hit to performance. Hope that helps someone who was leaving things relatively stock or maybe just undervolting with no change to power limit etc etc or playing with clock speeds on core vs memory.
  11. Holy hell, I need some more money for folding lmao. Teach me your ways, Sensei!
  12. Come to think of it, it would've been cool if I hadn't sold off all my mining hardware. Actually I'm realizing... I would've crushed his PPD. No kidding.
  13. What I've found is that setting clock speeds achieved with lowering power limit, to the same speeds without lowering power limit... Doesn't change power usage or folding power. If that makes sense. I'd strictly stick with lowering power limit and calling it good. Just from my limited experience so far though.
  14. I gotta' wonder... What the hell does LegendaryPoet have access to!? (top LTT folding member)
  15. So... If I were to guess, you could probably go into "Configure", then "Slots" and then remove your primary GPU? Maybe the one you have connected to your display? Not totally sure though, but it's a starting point.
  16. Yeah, see that's what bugged me, so it doesn't seem fully dependant on CPU cores at least. Are you seeing ~2x performance splitting your cores into two jobs, as opposed to the small gains using them all on one job?
  17. Yeah, the reported numbers are a bit weird. I'm not sure exactly how it comes up with that PPD number. Current work unit? Processing power over the last x-amount of minutes? I've found it hard to tell when changing a setting actually shows the correct number on the client. Also, that number is crazy low for a GT 1030. At least, I think it is? But it might just be that the full number hasn't shown yet.
  18. Really impressed by the RX 5700 series' performance from the numbers I've seen. I do wonder what a Fury or Radeon VII could do, but I'm not sure whether F@H can benefit from memory the way crypto mining for example can..
  19. Hey gang. I was just curious why F@H is so bad when it comes to multiple cores? I've got 12 cores / 24 threads and only get a bit less hashrate using say, 12 threads, as opposed to all 24, or even 20. Clock speed seems relatively unaffected too with those different settings, but the numbers are what they are, but with vastly less power consumption. Why? Could it maybe have to do with cache or something like that where there ends up being a big diminishing returns because you're already running close to max? Same kind of situation on my i7-4600u. 1 thread gets me like 5,000+ PPD. Two gets me like ~8,500. 3 gets about the same. And 4 actually seems lower! But clock speed and cache frequency do suffer after 3+ threads which is expected I guess.
  20. Well hey, with their processing power combined, you should still be working with some serious horsepower there! How many points per day are you seeing with that GT 1030? Just curious.
  21. Bigger capacity drives? That's probably the quickest fix I would think. I kind of wonder if it'd be easier to just use the Seagate 1TB as boot drive and mostly just ditch the 500GB drive? 'Cause you'd still have 6TB of space at that point. My "NAS" setup is just a Windows shared folder with one of those cheap Chinese USB 3.0 to SATA adapters. I think you're overcomplicating it. If you really want to use the 500GB HDD as boot drive, maybe just use a USB adapter to boot off that drive, and have all the others mounted correctly. Keep it simple. I think that should seriously solve your problem with the least amount of headaches possible.
  22. Gotcha. Restarting has worked, but not as of today. I'll try pausing for longer and then try to start folding again.
  23. Gahdamm, I can't get work units on my GPU with any regularity... The hell are you guys doing to get it to work? Also on my laptop, my CPU is deciding not to fold for whatever reason. Extremely frustrating.
×