Jump to content

Ratisbona

Member
  • Posts

    28
  • Joined

  • Last visited

Everything posted by Ratisbona

  1. Not even close. I did that experiment a few weeks ago on my 3900X. I downclocked it to about 3700-3800MHz fixed and undervolted it to 0.975V or something like that, to make it draw as close as I could get it to 65 W. And it still got over 2800 points in Cinebench. And from what I read about the 3950X, these chips are very much better binned that the 3900x again, so they should perform even better.
  2. I got a WU of project 14271 today, and it's producing about 1.500.000 PPD (Points per day) on my Vega. I've never seen points that high with any other work unit before, that's what I meant.
  3. https://foldingathome.org/faqs/fah-v7/v7-introduction/web-control/what-is-prcg/WHAT IS PRCG? This is an abbreviation corresponding to four numbers which identifies and distinguishes each Work Unit. The first number specifies the Project, which is the protein under study. A single project number is made up of thousands of individual Work Units. The remaining three numbers identify the specific part of the project that has been assigned exclusively to you. Work is rarely duplicated (except when a computer fails to complete an assignment in a timely manner) so every set of PRCG numbers is unique and only assigned to one person. Technical details of the PRCG numbers and how Work Units relate to each other are covered under the more advanced Simulation FAQ.
  4. Speaking of good Workunits for AMD: I got a 14271 for the first time today, seems pretty OP for my underclocked and undervolted Vega 64
  5. Yes, I've been getting only these 14196s for about two days now, I'm getting about 905k ppd on my Vega 64, while on the best WUs I hit 1.3M ppd. However still better than thed ~300k ppd I got with really bad WUs before so i'm not complaining, at least it's consistent
  6. This ist how I'm running my Vega64 (Sapphire Nitro+): +Memory on Stock Settings (950MHz). For gaming I'm clocking it a good bit higher esp. the HBM, but for folding this seems really good and I'm getting about 1,2-1,3 Mio. PPD (in good WUs...) at only about 140-150W and under 65°C:
  7. One of Germany's biggest electronics retailers (mindfactory.de) apparently just ogt a bigger shipment of these in, it's been available at mindfactory.de @retail since at least yesterday, that's longer that I've seen it stay available since launchday The sold units there went up from ~2800 at the beginning of last week to ~3500 right now as well. So shipments of these do come in.
  8. I don't use it primarily for gaming, my focus for it is on Videoencoding with Handbrake, so I'm running an all-core 4.2 GHz @ 1,225V which gives me a nice and stable around 75° on heavy all-core-loads. I just tested the stock behaviour for half an hour to see if it did anything, as I had never seen the advertised 4.6GHz reached on it before. By coincidence I had tested this same CPU a few days earlier on a X470 Taichi Ultimate Board which did boost the fastest core in Cinebench giving me a 205 points single core score, vs. 195 on my usual x570 ABBA-Board when boosting the "wrong" core. So it's a "measurable" performance-loss at the moment if that's consistent, but for gaming or any real world usage I doubt you'll notice it much.
  9. I've noticed the same behaviour with the ABBA-Agesa on my 3900x. It boost the fastest cores to 4.600 MHz or above (on idle mostly), but runs Cinebench on the slower CCD at about 4,3-4,4 GHz on stock BIOS. I don't care too much about it though, as i'm running an all-core overclock as well. This has been noticed by more tech-outlets as well, I think Tomshardware mentioned it, so it'll probably be adressed in the next update.
  10. Yes they should light up with the default Rainbow-effect. Maybe try reseating them in the slots?
  11. The best place to one initially in Europe was AMDs Onlineshop directly, they had stock in the first 2 weeks or so after release, but has since sold out. Mid-July the big german e-tailers (Mindfactory, Alternate,..) got some initial stock that sold nearly instantly with preorders. Since then they seem to get a few on a regular basis but that probably covers only the outstanding orders so "in-stock"-Stocks never seems to appear. But if you were to "join the queue" at one of these stores I'd feel confident to get one in 1-2 weeks.
  12. I have a bit of a wierd case as my 3900X seems to be quite a good chip, clocking 4.2 GHz at only 1.225V stable in prime without AVX, but at the same time having terrible thermals, preventing me to go any higher for longterm use. I'm certain at this point that the solder on at least to of the cores isn't great, so I'm getting quite extreme hotspots on at least two of the cores. In general I'm getting temps about 10°C too high (both with Thermal Grizzly Kryonaut and Carbonaut, so it's not the paste application at fault). But as 4.2GHz is "fast enough" for gaming (limited by GPU anyway) and better than stock for all-core-loads, I'm happy with this for the moment.
  13. You should consider the RX 5700 over a RTX 2060. Much better performance for the price.
  14. Changing the thermal paste CAN make the difference. If it's dryed up and cracking, that can lead to temps like you're observing. No guarantees though , but it's the first thing I would check and change.
  15. If you want to go as cheap as possible I'd try first to keep the current motherboard and RAM. With themost current BIOS this does support the 3700X and with the much improved memory controller of 3rd Gen Ryzen your RAM will most likely work just fine as well. First gen Ryzen was notoriously picky with RAM so no surprise that you could not get it to run 3200 MT/s but that's much better now. For cooling I can recommend something like the bequiet Dark Rock 4, at least here that a good bit cheaper that the Noctuas and cools my 3900X just fine so won't have trouble with the 3700X either
  16. I didn't say it was better, what I meant was it's usually better for gaming performance to allocate more budget towards the graphics card and less to the Mainboard than you did in your partpicker list. As such a B450-board is better value (not "better performance") than your x470. Vexicus recommended an eaven cheaper one, have a look if that fullfills all your needs. If you really need more USB3.0-Headers (or even 3.1 Type C) for you'll still need a better equipped x470-board. A SSD can use either a SATA 3-Port or a M.2-Slot. Both will be present on pretty much all AM4 Boards so you shouldn't have to worry about it too much. Do you know what modell you have?
  17. I'd knock 40-50$ off of the Motherboard going for something linke a MSI B450 Tomahawk and add that Budget towards a better Graphics Card. A X470 Board for a Ryzen 2600 is slightly overkill.
  18. Running stock will most likely even be better for gaming as it allows the CPU to boost to higher frequencies on a few cores than you can expect to reach with any overclock (on "normal" cooling). As many of the current games still scale more with frequency than threadcount the higher boost clocks on a stock Zen2-CPU often help.
  19. Same here, I have the x570 Taichi, which has quite an annoying sounding fan on the Chipset. But I simply changed the fan curve to only ramp up when it hits about 80°, and I've never heard it since. In normal use the temps sit around 60° under gaming up to 70, but absolutely not meling the board.
  20. No. As AMD managed to raise the IPC (Instructions Per Clock) by about 13-15%, this means with the same clocks the newer 3600 is about 13-15% faster than the 2600. Plus 4.2 GHz on Ryzen 5 2600 is really not easy and depending on your Chip might prove impossible without extreme cooling. Wether the about 25% out of the box performance increase is worth the increaed price for you only you can decide. But you will get more performance with the Ryzen 3600
  21. Ok to bring this back to topic: One of the most commonly used tools for this is "HandBrake". You'll find lots of tutorials for this on Youtube. It's got a lot of options to play around with, this should fit your needs.
  22. My guess would be your SSD is too full to run the sequential test. That's why it can't complete. The Sequ-test can use up to 100GB on a fast drive like yours
  23. I can recommend the Dark Rock 4 (Non-Pro), that just about clears the second RAM-Slot (on my Board, YMMV) so you're unrestricted in Height for the RAM, as long as you're only using 2 DIMMs. Plus you can actually see the RGB-RAM and it doens't vanish underneath the giant cooler
  24. The Intel drive is using cheaper and worse memory technology, that's why it's cheaper, but also not as good as Samsungs drive. For more details watch this video:
  25. I think an easier explanation would be to look a how many possiblilites you have for winning field combinations, knowing that green is not a winning field. Given the cinstraint that the two winning fields have to be next to each other you have four options: Winning fields could be blue/purple, purple/red, red/orange or orange/yellow. As you are determined to only be allowed to shift to yellow, you'll have a one in four chance of that being in the winning combination. Spinning again gives you a two in six or one in three chance, which should hence be the option of choice.
×