On 10/15/2018 at 12:27 AM, suzuki said:

One from me,other will follow,i hope.

 

Suzuki- 8m41s922- i7 990x@ 3991 MHz, NB @ 4540 Mhz, EVGA X58 Classified E770, 3 x 2GB Supertalent  @ 936 6-6-5-18-52 1T

 

Nice score man !
Are these Elpida Hyper starting in 6 6 5 18 1t trfc=52 ?

I have been through ddr2 d9gmh and ddr3 psc or samsung, but have no clue for elpida hyper how much i can tighten timings, secondaries tertiaries in particular...

Cheers

Not sure what you mean with “starting”.

The consacrated timing formula is 6-7-5 or 7-7-6 and sky is the limit. There is also 7-8-7 on which some may work better. Up to 1750 mhz i used even 5-6-5 :). Hyper is awesome !

1 hour ago, suzuki said:

Not sure what you mean with “starting”.

The consacrated timing formula is 6-7-5 or 7-7-6 and sky is the limit. There is also 7-8-7 on which some may work better. Up to 1750 mhz i used even 5-6-5 :). Hyper is awesome !

Ever tried going with 5-6-4 or similar? Also considered trying to find some interesting BBSE? I have some sticks that can do 5-8-5 at 2000 MHz at ambient given the right CPU+board combination (excuse the garbage efficiency, wasn‘t running an optimised OS for that, just testing if the sticks could handle 32m at those timings without crashes, sadly didn‘t get around tuning subtimings before the CPU died in a board accident)

I would be very cautious with timings like cas5/Cas4 etc on certain platforms, on sandy bridge for example I tested this long years ago only to find that this was not real.. and it was confirmed several times by others as well - also wcl settings for example on different platforms can influence if timings are real or not, not all BIOS shows and you see at cpuz is real

I have one of the best kits of bbse arround ,i think . It can do easy 2700 8-11-7 spi32m all tight . Never tryed them instead of hyper cause i know untill sandy hyper was the deal and no other kit of different ic’s could beat it . Might give it a try after i relocate as now there is no time to do anything.

 

I remember a post from @nachtfalke on a different forum where he modded a cheap Asrock and it was a beast with timings up to 1800 mhz ,need to find the results. Maybe it is also board related ,who knows....

Edited October 27, 20186 yr by suzuki

3 minutes ago, websmile said:

I would be very cautious with timings like cas5/Cas4 etc on certain platforms, on sandy bridge for example I tested this long years ago only to find that this was not real.. and it was confirmed several times by others as well - also wcl settings for example on different platforms can influence if timings are real or not, not all BIOS shows and you see at cpuz is real

Sticks run 5-8-5 at 1.84V, not any less, runs 6-8-6 easy at 1.64V. Also some score difference between the two. 

Then it should be OK, but you won´t beat Hypers nontheless because of trcd limit.  1,64v for 2000c6 is not that bad, maybe test them for samdy bridge or ivy if they can do 2600 8-11-7, this makes a lot more sense.

1M @ intel E2140/air

ASRock X48 TurboTwins-WiFi
-cpu vcore -VID mod add +300mv ( 1.6permis de bios -1,568 real)
-cpu vdroop mod
-cap mod( 4 tantali back pcb -330Uf)

-ram vmod / solid cap mod
-NB vmod +150mv ( 1.370v( 1,27default voltage) max permis de bios( aka Highest denumit in bios )
1xSTT2000cl7

 

Look at samsung power,they were done by nachtfalke. You have also my screen with 5-6-5 ,full timings.

 

I leave you a classic example for my point that being cautious is never wrong - one value changed as cas was set in Bios to 5, voltage same, rtl same and time same = no real cas 5

On 10/15/2018 at 12:59 PM, suzuki said:

I am missing an option from bios on classy which it is on Asus . I estimated at least a 3 s improvement but 36” seems doable if you have everything right ,like ram on ln2.

Got a Improved score yet suzuki? 

No ,i am moving out of country so everrything is packed,after 15'th of nov will have some time to settle,if everything comes on time.

5G  Thuban , will run it up again later

Here's a couple with 4770K & 4 DIMM Sammy D-Die.

Haswell 4g

Edited April 5, 20205 yr by Bullant

On 4/5/2020 at 8:29 AM, Bullant said:

Haswell 4g

That was insane, Bull... Z97X SOC Force LN2?

8 hours ago, ivaneza said:

That was insane, Bull... Z97X SOC Force LN2?

Hey mate, long time....yeah Ln2 board, hope you are well 

4GHz 45nm; not maxed yet. Refusing over 4GHz uncore since I don't think that plays within the spirit of this challenge.

 

https://hwbot.org/submission/5146492_ 4GHz Westmere. Still playing with only 4GHz uncore. Lazy mems since my chiller isn’t set up right now. Also very sloppy final loop, just a rerun could probably get this one down half a second or more.