Jump to content
HWBOT Community Forums

Leaderboard

Popular Content

Showing content with the highest reputation since 09/18/21 in Posts

  1. Treat it as it is, 16C is still 16C doesnt matter slower core, faster core, or different clocked between core, core count is still same. If 12900K/KF lose to 5950X/10960X, leave it as it is, because actual multithreading performance is slower, they still have single core performance advantages anyway. No need more complicated rules, now is already complicated enough.
    6 points
  2. I don't see any rule change. *shrug* Been the same since it started.
    4 points
  3. In my humble opinion i would go with : ( We simply list the 12900K as a 16C CPU. Might sound like an easy option for now but I see that this would make it very difficult for the future years especially thinking about that AMD will eventually also use different performing cores on one single chip. ) . Then wait and see how it goes from there. Best bet is to wait for the cpu to be out and available and see where performance lands , then we can make a propper informed desision. No need to rush things or get ahead of it ,let's wait and see , meanwhile listed for what it is a 16 core cpu.
    3 points
  4. If 12900k has 16 cores, treat it as a 16-core cpu. Period. If Intel decides to make 8 of them stronger and 8 of them weaker, have HT on one and not on the other, thats then Intel's decision. From what you hear it even performs like a "true" 16-core 5950x. So what's the big deal. Also 12900k is monolithic. For example the 5950x isn't even monolithic and has chiplets. Can I disable chiplets and make it join the 8- or 4-core category? No I can't. So if you have a 12900k you can join two categories and essentially gain double the points in some way, that's just weird and complicated. Allowing disabling part of the chip so it can joins a different category so it looks stronger is contradicting previously made statement regarding disabling cores for frequency validation. ppl can disable all they want it is still an 16-core chip, and therefore it should get treated that way eventhough some cores are stronger than others, some are more efficient than others. It's the overall performance that counts. Regardless of the amount of threads right? When Intel comes out with with quad HT, what do we do then? It's still X amount of cores but 4-times the amount of threads. The way I look at it is the individual cores just got more powerful/efficient. But plz don't make it more complicated with dividing CPUs into multiple categories. Please correct me if I'm wrong but we do we differ between a 6-core with HT and one without?
    3 points
  5. First off congrats on that run, it's a good one. Suggestion: You should live stream those attempts at the records you're after - Including mine if you wanna go for them. 👍
    2 points
  6. Cool! Shall be giving my 8320 another run out very soon, so be interesting to see how we get on
    2 points
  7. 32m is challenging, but isn't that why we still love it. Take this picture below, this is TestMem5 Extreme1@anta777.cfg, HCI MemTest 7.0 and LinX 0.9.11 stable but I can't pass 32m with these settings. It needs higher DRam V, IO Mem V and SA V to pass
    2 points
  8. DDR4-5600 19-27-27 1T Very Tight passed
    2 points
  9. WB edition is nice for OC, since you dont have 2 million screws to remove like air cooler Elmor is working hard on a direct voltage control for 6900XT fingers crossed if you dont have an EVC yet def grab one.
    2 points
  10. XFX Speedster ZERO AMD Radeon™ RX 6900XT RGB EKWB Waterblock Limited Edition - New vga based on the xtxh chip. Clocked at 2525/2000Mhz EK FC H2O Block. 14 + 2 Phases. https://www.guru3d.com/news-story/xfx-now-offers-radeon-rx-6900-xt-speedster-zero-wb-graphics-card.html
    2 points
  11. ERam install for Super Pi 32m XP/32 - Imgur
    2 points
  12. Just use auto unless it has issues training. If it has issues training, then try each of the settings until you find one that works. On the older bios I had to mess with them. On later bios, auto works just fine for me. Apparently there is a way you can add additional texts to windows and then it works. I do it the lazy way. My was is just to run the normal AFD installer that you would use on w10. It won't work and it will give errors when installing. After installing full version, the file that is inside that xp folder will show texts and work correctly. No need to use the version that was installed via full installer.
    2 points
  13. Tried random UD3LR with my 720 fsb E8500 today, seems to run alright https://hwbot.org/submission/4819935_ground1556_reference_frequency_ga_ep45_ud3lr_710.36_mhz
    2 points
  14. Joining again with new memory, much improved quality vs previous all DJR that i got.
    2 points
  15. Nice , awesome set up , great score bro well done ! I see good 3d scores i instantly hit like :D
    1 point
  16. 1 point
  17. Sorry, yeah so you want to stay 1:1 fabric and memory. Safe or not I'm not sure but on air I have used 1.45vcore and 1.45vsoc. dram 1.92-1.96 for 4800 14-14 Sounds like your voltage just needs to be increased.
    1 point
  18. Hey guys, let's talk and post results of the hynix djr ic here. You already know that these broke memory frequency WR on Z490, well things only got better on Z590. With the better IMC from RocketLake now these sticks can be used for competitive benching as well. I had a shot with them on Asus Maximus XIII Apex and things are interesting. Let me put it this way. On air a decent Hynix DJR Kit is most of the times better than a very good B-die kit. On cold if your Cpu can handle both high core clock and memory clock at same time, a decent Hynix DJR kit is similar to a great B-die kit. To tell the whole story, Hynix DJR is way easier to handle, no need for maxmem, if it enters windows it usually runs without hickups. I was really excited about the Hynix DJR performance and really hoped that they will send the Samsung B-die at the thrash bin after so many years of painfull benching, sadly they cannot top but they are a very very viable alternative. You can find 3 profiles in Apex Bios which work and they give a good starting point if you want to try Hynix DJR on your brand new Apex and Rocket Lake Results on Intel 11900KF and Asus Maximus XII Apex: Hynix DJR cooled down to -150: DDR4-5866 17-27-27, memory score at 5 ghz on older microcode bios ==> 12058 Which is better than the memory score of Samsung B-die DDR4-5333 15-15-15 (12034) in same conditions. Hynix DJR on air can do between DDR4-5600-5700 with C19 and the performance is equivalent to a Samsung B-die DDR4-5000-5066 C14. Cold findings with Hynix DJR: - they scale good till -50, at -150 there is a small gain but not much. - trcd and trp remain same air vs cold, no gain - trfc which really helps these ( 20 values tighter in TRFC gives 100 points boost in geek3 memscore) sadly is the same on cold - secondary and tertiary go very low, on ln2 tighter than air but no performance gains. There are no twr or trtp or whatever issues Samsung B-die has. Here is my best result with HyperX DDR4-4600 C19 based on Hynix DJR on ln2: Similar performance to Samsung B-die at DDR4-5333 15-15-15 Feel free to post your results and experiences as well, maybe we can find some better binns or better sorting methods.
    1 point
  19. @shar00750 no core unlock possible at all on this chip. 2 out of 7 560BEs unlocked for me
    1 point
  20. I used 3c 3t because my strongest core was not 4/5 and settings between both runs were identical except the os having a larger maxmem set on the latter and retiring my Raptor. tRAS lower than 21 will not pass on my pleb tier dual rank kit and lower tRFC doesn't pass consistently. As for atypical tertiaries I can tell you I'm much looser than what have posted on a number of them. Sometimes tighter doesn't equal faster and most definitely not if you can't finish 32M. RTL game is strong for sure because . . . OCF I am using 1.5v on an NH-D15 with 75f ambient temp so I don't think there's any gas in the tank in this regard. Cache higher than 48 fails to boot nor have any semblance of stability so mega fail in this dept. I'm also Gear 1 limited (and or memory limited) on this cpu as 3866 can't train here unless I use < 100 bclk so there's no going higher freq wise atm.
    1 point
  21. 1 point
  22. You're not the only one who suffers from this CPU-Z Multi-Bug... absolutely annoying.
    1 point
  23. No delid, just LM on IHS. I should pop the top and try delid huh? We all know there are 5.5 subs coming, haha, so maybe I wont risk killing this chip...
    1 point
  24. They're wrong, all reviews state 125/143. Fixed. I'll check my cards later this year. When GPU-Z or other utilities will tell all of them apart.
    1 point
  25. Nice! I'm considering LM earlier but didn't do it because my block is pure copper and LM will stain it - maybe I'd find some nickel-plated blocks and try again with LM. Keep pushing it, cheers
    1 point
  26. Thank you! Yeah, it's big water and liquid metal!! LM is magic and my ambient temps are not bad even though it's summer.
    1 point
  27. It seems like i managed to accidentally unlock the RAM and NB dividers on a Xeon E5504, so far i only managed to reproduce this with a very specific OS and once you completetly shut down the system it's locked again (hotswapping ssds is possible though to get extra mem speed on diffrent OS)... Here's some 32m runs with 2:10, 2:8 and 2:6 to prove that it is not only display bug of CPU-Z Will upload video of how i did it later, maybe even immage of the speciffic OS i used, if i can't find what exactly causes it... How I did it: 1: take windows 7 (so far only one of my copys worked) 2: reset during boot, so you get the option to boot in safe mode 3: set memory settings that should work but are locked in bios and apply 4: when windows asks you coose boot normaly and hope it locks up 5: press reset button on board (do NOT press powerbutton) 6: POST screen should show the ram/nb settings you applied 7: boot into OS, if you wan't to bench on XP or simmilar disconnect SSD while in OS and press reset button and it should boot into whatever OS you connected with the propper settings Hope someone will be able to reproduce this EDIT: Youtube video: I have now tried this on X58A-OC and X58A-UD7 without any success, so probbably only works on ASUS Boards, unfortunately i only have the one Rampage III Extreme, allso i managed to get it to work on all windows 7 installs i have, key element was to get the recovery screen where the cursor is on start "automatic repair", not on "start normaly"
    1 point
  28. Hi, looking for a fully working 2x2GB kit either Corsair Dominator GT 1866C7 or 2000C8 ver2.1 Thanks a lot
    1 point
  29. Bit of fun but time to warm it up , put MOBO in oven and relax 🙂 My fingers are freezing lol. I don't like the clean up afterwards , pain . That was only 20L 😂 X265 really goes through this stuff.
    1 point
  30. I kinda like how 1x, 3x and 5x are "legacy" global so to speak, but that's a bit off topic...
    1 point
  31. If you call it a 8 core its is fighting against 5800x and 11900k IMO that doesnt seem fair at all. If you were to pickup the phone and say "hello MR.Intel, how many cores does a 12900k have?" they would say it has 16. In my opinion its a 16 core processor, this is due to the fact that it has 16 cores. We need to not dilute the rankings by catering to hopeful companies to claim PR victories.
    1 point
  32. Woah cool! So why aren't they on the official page? That's where I go cos I know it ain't broke! 💔
    1 point
  33. You gotta check ozzies quote, he properly quoted them
    1 point
  34. ddr overvolting on cypress gpu arhitecture will not help the mem overclocking at all in 90% of cases, the problem is the mem controller inside of gpu , hd 5970 use Hynix T2C ( we find that ic´s on HD5770 aswell ..and they ran with a good MC over 1500mhz.. so there are no limitation from ram aswell)they can ran at 1400+mhz without any voltage ajustament! so in all situation MC will be responsable for this..... ram vrm is also volterra in one phase for EACH gpu gpu´s regulator: VT1165mf ( conected pin9 to GND with 50ohm trimmer)drives three power MOSFETs(VT 1157SF) with a PWM signal for each GPU so you have separatly voltage control 3+3 for gpu´s...also software hack allow controll ajustament for VT1165MF without any problem ,but for me hard mod will be always best way from far.. for vmem don´t need at all a pwm chip to drivers only one phase with VT237WF integrated switching regulator 34A , 1.521v default...because it is a digital power mosfet, so that power mosfets will do alone this job als power regulator and pwm aswell! for powered uncore I/O it is also 2phase volterra with 2xVT1157SF power mosfets and one Vt1165mf pwm controller ... so in totaly we have 3 x VT1165mf pwm´s ... 2 ( 3+3 phase for each gpu)are for signal GPU regulators and another one is for uncore I/O (2 phase) for RAM we had 1+1 phase VT 237WF for each gpu , power mosfet and pwm aswell...
    1 point
  35. http://abload.de/img/asus8ps8o.png http://abload.de/img/msipfs93.png http://abload.de/img/1400c9_215v_32m_z87roowk1z.png However, I'm not sure if 1400 9-12-12 is 100% reliable on this kit.
    1 point
  36. Hi, after a major test marathon last months I have a couple of kits left which might be useful for benchers . All tests screened were done on air on either M5Extreme or M5Gene Z77 with 3770K OEM cpu Samsung 1. 2x4GB GSkill TridentX 2400 C9-11-11-31 1,65v Mint condition, box and fan are part of the deal (may raise shipping cost). Tested for 2800 9-12-12-18 112 (TWCL6, LB6, TRRD4) 1,99V 32m on M5E, voltage not optimized, 2800 10-12-12-25 medium samsung profile at 1,69v and 2600 9-11-11-21 tight samsung profile twcl6 at 1,73V. High trcd limit, no problem to boot and 1m 2933 10-12-12. SN 1500, most likely D-HCH) or HCK0 2. 2x4GB GSKill TridentX 2600 c10-12-12-31 1,65v Mint condition, box and fan are part of the deal (may raise shipping cost). Easily do 2600 9-11-11-21 around 1,71v, tested also for 2800 9-12-12-21 tight TWCL6 with 1,96v 32m on Z77. Excellent mems, better for example than several 2666c10 kits of different manufacturers I tested. High trcd limit, 2933 10-12-12 no problem boot and 1m at acceptable voltage. SN 1500, most likely 2gbit D-HCH9 or HCK0. 3. 2x4GB GSKill TridentX 2666c10-12-12-31 1,65v Like new, come in box with unused fan. Tested for 2800 10-12-12-25 1,65v, 2600 9-11-11-21 1,7v and 2800 9-12-12-21 1,92v 32M on MVGene and Extreme. 2933 10-12-12 below 1,8v boot and 1m. SN 2500, most likely Samsung 2Gbit D-HYK0 PSC 4. 3x2GB GSKill Trident 2000 9-9-9-24 1,60V Optic mint, technical flawless, come with original GSkill Blister packagaing. Very high TRCD-limit, all three sticks do 1200 7-10 32M. Kit is a bit uneven in performance, best stick does 1200 7-10-7-28 ~1,68, second one 1200 7-10-8-28 1,69V and third one needs 1,73v for 1200 7-10-7-28. Tested two sticks for 2600 8-12-9-28 Hard secondaries and extreme thirds for 32m at 1,89V, 8-12-8 should work as well. Can go because I have better PSC, and I see a very good chance these are good performers on cold, which I don´t use. SN 10221040, XPSC 5. 2x2GB GSKill Ripjaw 1600 7-8-7-24 1,6v GBRM X-PSC, tested as kit for 1200 7-11-8-28 1,66v, 1250 7-11-9-28 1,72v 32M on GeneV. Latest tests were 1300 8-12-9-28 auto 1,8v, 1300 8-12-9-28 TWCL6 LB6 1,85v and 1300 8-12-8-28 hard secondaries/thirds 1,86v, not optimized on voltage. Better stick is a winner, 1300 7-12-7-30 easy at 1,75v auto on extreme, needs around 1,8v for TWCL6 hard subs. The 1300 8-12-9 are memtest stable, if 8-12-8 works depends, as always, on your system. Was offered before, now again at higher price, sometimes it is cheaper to trust a word than to wait for the screen . In original GSKill Blister packaging, early 1040, XDX PSC Prices (without shipping) 1. 115 euros excl. 2. 150 Euros excl. sold 3. 180 Euros excl. sold 4. 65 Euros excl. sold 5. 75 Euros excl. sold Best regards websmile Usual note, no returns; no warranty or guarantee provided because this is a private sale and I deny this, if you don´t accept terms, don´t buy. OC-results, if provided, are offered to show the potential of the hardware and results may vary.
    1 point
×
×
  • Create New...