Don_Dan

Are the Kingston MFR rated 3333 CL16 or 3300 CL16? Are they already sold to rivaldokfc? If not, please consider them reserved.

If Alex doesn't take both Samsung ES kits I'll take one.

I'll take two of each.

You're welcome! Some of my own tests that I did two years ago. E8400 7x 572MHz, Biostar TPower I45, 2x1GB Ballistix 1:1 4-4-4-12 2T PL9

Nice work!

Stark!

[Guide] Asus P5B Deluxe / Commando P965 JP Pmem Strap Tweaks youngpro posted this in the above thread. FED1403Fh => 23 Biostar Tpower P45 first look with Team.AU youngpro posted some more info in his thread about the Biostar TPower I45. FED14030h => 06 FED14038h => 26 OPB 2008週末極é™è¬›å ‚; fsb651é”æˆ-P45çš„MCHBAR Tweak探索 OPB shared some settings for Biostar P45 in this thread. FED1403Fh => 22 Some QX9650 results with MSI P35D3 Platinum, STT DDR3 1600 and 2-Stage Cascade before shared some info in his thread about the MSI P35D3. Any ideas about these X38 MCHBAR registers? before, FELIX and The_Stilt shared lots of info on different registers in this thread. Asus Rampage Extreme: having fun with MCH, Spi under Air and Cascade coolings before shared lots of info in his thread about the Rampage Extreme. Thanks. Well, MemoryControllerHub registers fall within FED14000h to FED14FFFh (size = 4096). FED14030h is just an offset among tons of others (effect of the sequence from FED14030h to FED14043h on memory bandwidth has been firstly introduced by Jean-Philippe aka "JP" @IxtremTek.com, also registered as "Learn" here @XS), and it remains a bit unclear to what it really stands for. It's strap dependent on P965 but linked with other options on more recent chipset such as P45, X38 or X48. It has to do with the following BIOS options on DFI boards: Enhanced Data transmitting (EDt) and Enhanced Addressing (EA). These are chipset timings; generally speaking, changing offsets here may need more or less vMCH. About FED14200h, it's the beginning offset for the first DRAM channel timings register. Among this one, one could find all DRAM timings that are also used by Memset (tCL, PL, etc.). Brother register is located at FED14600h and obviously commands the second DRAM channel timings. Generally speaking, changing offsets here may need more or less vDRAM. Although Memset allows a bit tunning there, other timings can be changed by editing MCHBAR: Read Delay Phase adjust, Write to Read Delay Same Rank, Write to Read Delay Different Rank, Read to Read Delay Same Rank, etc. Those timings may have a positive to very positive impact on bandwidth. Sometimes, tightening some of them may need to increase PL... Basically, they are more or less interdependent. Then, FED14C50h to FED14C57h is a sequence of the memory sizing register. Hard coding is automatically done accordingly to DRAM to FSB ratio. To my knowledge setting that results in better bandwidth is FF FF FF FF 00 00 FF FF (8bits). It's in fact the default coding that is used for 1:1 ratio. Enhancement depends on overall coding indeed; I eventually tried it on MSI P35D3, and improvement was extremely good... but on Rampage Extreme, improvement isn't that impressive, but still here. vMCH is involved in this job. Concerning which values to use; well, as you may know coding is done using hexadecimal. Here's the correspondence to decimal: 0 0 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 A 10 B 11 C 12 D 13 E 14 F 15 So when changing timings, you generally have to think about clocks behind hexadecimal values. Commonly observed logic is the lower the number of clocks, the faster the performance (this is not always true, but you have to see that by yourself) For instance, here's what I was guessing months ago about FED1403Fh: Well no. FED14C50h is in my humble opinion a bit different, so let me take another example. FED1403Fh is an address which corresponds to 'possibly' 2 timings. These timings 'could be' CC, FF, 00, or whatever within 0 to F individually. When setting FED1403Fh to CC, we could reasonability think that the 2 timings are each set to 12 clocks. Addresses like FED1403Fh or FED14C50h aren't chosen randomly but correspond to language machine locations for memory controller hub functioning instructions. Some are available thx to intel datasheets but others have been discovered thx to hours of intensive researches : Live MB OC Report :: ASUS Rampage Extreme Some info on skews in this thread.

This thread is a copy from XS. After the hack I was afraid that this thread was completely lost, but @FUGGER was kind enough to check another back-up for me and I found the first two posts of the thread in there. I decided to share this thread with all of you to save it from being lost forever. ______________ Recently I became interested in strap tweaking again. I had only used some of them on the Commando in the past, but now I want to try them on a newer chipset. Unfortunately many of the old threads are either gone or screenshots are missing. Therefore I decided to collect all available info in this thread and copy all the screenshots to my personal imgur account. The first post will be about strap tweaking on P965, the second post will be about strap tweaking on P35/P45/X38/X48 and the third and fourth post will be about my own testing. Histoire de strap Strap tweaks were first discovered by JP on the french forum Ixtremtek on the thread Histoire de strap, which is only available on the web archive nowadays. Unfortunately only the first page of the thread is available, so I'll just copy the first post. ...tout bon: Le registre 8bit est bien en C00h Donc 20h pour toi. 20h->0.010.0.000b 0.(010).0.{000}b La valeur entre parenth?se et le ratio m?moire, La valeur entre crochet est le fsb strap. Je suis sur la bonne route j?ai donc fait beaucoup de dumps de la mchbar ? diff?rente fr?quence et c?fficient m?moire dont voila un aper?u : http://www.overclocking-jp.eu/public...aps-ratios.png Ont peu donc voir qu?il y a trois valeurs pour le strap qui doivent ?tre (je ne serais pas affirmatif n?ayant pas la documentation confidentiel du P965?) 533, 800 et 1066. Si vous vous demand? pourquoi une valeur pour le ratio m?moire peu correspondre ? un m?me ratio c?est parce qu?il y a un changement de strap qui autorise certain ratios, ce doit ?tre ? priori comme sur d?ancien chipsets intel, les I875 / I865 en autre, ces valeurs qui sont document?? http://www.overclocking-jp.eu/public...atios/P865.png Il d'ailleurs un (vieille) article mais tr?s int?ressant pour ceux que ?a int?resserais sur x86-secret en rapport avec le sujet, dommage qu'il n'y en n'ai plus... http://www.x86-secret.com/popups/printarticle.php?id=87 Vous n??tes pas sans savoir qu?il y a une diff?rence de performance m?moire quand ont boot ? 400MHz ou 401MHz, fr?quence o? il n?y a pas de changement de strap. J?ai donc cherch? ? savoir comment rattraper ces diff?rences de perf m?moire et je suis tomb? sur le jackpot LOL J?avais d?j? un peu boss? sur les diff?rences qu?il y avait en bootant ? 400 et > Voila les diff?rences, celles int?ressante que j?ai pu trouv? jusqu'? pr?sent : http://www.overclocking-jp.eu/public...hbar_part1.png Je me suis fait une petite s?ance de bench m?moire sous Everest 3.80.948 version payante : J?ai appel? les groupes R1 R2 R3 R4 et R5 changement fait en mode 32bit. La configuration : P5B deluxe C1 Bios 1101 E6600 Le syst?me de refroidissement d?origine du chipset ? ?t? retir?, pour le remplacer par celui d?une carte m?re DFI NF4, ventilateur coup?. 2x512MB de Kingston Timings 4-4-4-12-6-35-10-10-10-10-D lien vers une photo: http://www.overclocking-jp.eu/public...s/000_1488.jpg Boot ? 401 x 9 puis passage ? 400 sous windows : http://www.overclocking-jp.eu/public...4009-stock.png Avec le tweak R5 http://www.overclocking-jp.eu/public...19_4009-R5.png Avec le tweak R5 et R4 http://www.overclocking-jp.eu/public...9_4009R5R4.png Avec le tweak R5, R4 et R3 http://www.overclocking-jp.eu/public...4009R5R4R3.png Avec le tweak R5, R4, R3 et R2 http://www.overclocking-jp.eu/public...09R5R4R3R2.png Avec le tweak R5, R4, R3, R2 et R1 http://www.overclocking-jp.eu/public...R5R4R3R2R1.png Boot ? 400 x 9 : http://www.overclocking-jp.eu/public...400x9stock.png Une fois tout les tweaks appliqu? ont peu s?apercevoir que les performances sous everest sont tr?s proche du boot ? 400MHz. Apr?s un court test de monter en fsb avec boot ? 400 et boot ? 401 plus les 5 tweaks je suis arriv? ? une m?me fr?quence, avec 1.45v au MCH, 430MHz. Voila pour le moment mes recherches toutes fra?che sur le sujet, j?ai fait d?autre d?couverte pas tr?s importante ? ce jour mais je vais s?rement jeter un ?il du cot? des dumps fait sur P5K. Un grand merci ? F?lix, j?esp?re ne pas avoir fait de b?tise en postant mon MP ? Boblemagnifique bien s?r et sans oublier Franck (de cpu-z). Commentaires bienvenus [Guide] Asus P5B Deluxe / Commando P965 JP Pmem Strap Tweaks Based on the above thread eva2000 wrote a nice guide, I'll copy the most important posts. E8500 6GHz Benchable Club NewBeetle found some very interesting settings he posted in this thread. STRAP-мод Topmods posted an article about strap tweaking. Since he explaines the process very thorougly, I'll just summarize his settings. The testing was done at 400MHz and 500MHz FSB. FED14030h => 06 (was 00) FED14034h => 10 (was 00) FED1403Eh => 03 (was 2B) FED1403Fh => 24 (was 84) FED14040h => 09 (was 0B) FED14030h => 06 (was 00) FED14034h => 10 (was 00) FED1403Eh => 03 (was 2B) FED1403Fh => 23 (was 84) FED14040h => 09 (was 0B) FED14030h => 06 (was 00) FED14032h => 0F (was 02) FED14034h => 10 (was 00) FED1403Bh => 37 (was 00) FED1403Ch => 17 (was 07) FED1403Eh => 03 (was 2B) FED1403Fh => 24 (was 84) FED14040h => 09 (was 0B) FED14030h => 06 (was 00) FED14032h => 0F (was 02) FED14034h => 10 (was 00) FED1403Bh => 37 (was 00) FED1403Ch => 17 (was 07) FED1403Eh => 03 (was 2B) FED1403Fh => 23 (was 84) (needs high Vnb ~1.77V) FED14040h => 09 (was 0B) A little comparison at high FSB clock to show the difference: Miscellaneous No idea where this screenshot came from, but I found it on my HDD.

Closed.

I'm sorry, but since I never had luck with GeIL in the past and since this kit doesn't convince me either, I'm not interested in it anymore. In the meantime I found a kit of Corsair Dominator Platinum Orange with results from both X99 and Z170, which should satisfy my needs.

I'm looking for a good pretested kit, so the price will depend on the kit I can find. Is it the kit as shown here? I've read a few reviews and all of them have trouble loading the XMP or getting the kit to work at the specified settings. On THG it says that the kit is only certified for Gigabyte’s X99 SOC Champion and MSI’s X99S XPower AC. Did you test only on X99 or also on Z170? I haven't found any pictures of the kit without heatspreaders, but I think the specification of DDR4-3400 16-18-18-36 points at Samsung D-die.

Hey guys, I'm looking for a decent kit with Samsung K4A4G085WD-BCPB, a.k.a. D-die, ICs. dual or quad kit high bin kit (3000 CL15, 3300/3400 CL16 etc.) flawless optical and technical condition original packaging Must ship to Germany, payment via Paypal.

Thanks for the heads-up, Charles.

Fixed!

Congratulations for being the first one to reach DDR4-5000!

I take #1!

Congrats, riska, amazing score!

Is the P55 Classified still available? If yes -> PM!

Here's a MHz. And a like.

Well done!

Congrats for this achievement!

My package with 1), 2) and 10) arrived on Monday. Thank you very much!

I'll take 1) nF4 SLI-DR, 2) nF3 250GB and 10) A-Data D9. 90€ shipped? Please send me a PM!

I think we can all agree that KP's scores are legit. I'll close this thread before it gets out of hand.