**Official DFI LanParty UT 790FX-M2R Review/Overclock/Guide Thread**

[KTE]

thats not the problem, trc 11 runs fine even with ddr1066, but i'd like to try above 26 to reach 1100mhz, which should be possible with powerchip ic's but they will need higher trc then 26, in epp-profile it says 41 @ ddr1000

Are you sure they'll need such high tRC? Usually they don't [ignore the rated specs]

[Oese]

maybe, mainly i think so as well, but i slightly miss the possibility to try...

perhaps my rams are just at their limit, but whats surprising to me is that they hit 1066 all at stock, even at tighter timings, but there seems to be no way to tweak them to just 10mhz higher, tried voltage, timings, drive strength...

but drive strength i'm not really at the end of testing as there are that many combinations and without any clue its just trial and error...

maybe i'll try some higher clock drive strength soon as Tony and someone else suggested over here..

perhaps its even the imc of my chip.. will try further when my 6400+ arrives..

[justapost]

Achim are you getting a B3 anytime?

I planned to wait for 45nm but how long will it take till they are release? At least half a year I guess. And then i'd end up with an early 45nm stepping. So I skipped first plan and ordered one yesterday, should be here mid next week. Also expecting an Gigabyte 780G mobo here on ~monday.

Good thing about AMD is, their platform is extremely tweakable because their full arch BIOS gude is freely available to all. With Intel, they only give it under NDA and so no one but a few knows facts but many assume false conjectures

What NDA! So that is why none in the intel section could tell me how C1E and EIST p-states work on intel. If i'd know this before id have taken a look into the linux kernel source first.

[Undersea]

The 9550 has shown up on newegg.

Anyone see where 9850 are available yet in the US.

Brother I plan on doing this phenom learning with you, I am at a disadvantage because I don't understand many of the bios options including memory settings outside of the very basics like CL, tRCD, tRD & Command Rate.

I've checked Google, Clubit, Zipzoom & Directron also.

[Oese]

oh you will understand only by experience... and the problem is, theres something new with every platform you use.. so real professional understanding (when this is possible with this kind of matter) will require lots of testing different ram and settings...

[KTE]

Ah ok Achim. Yeah I'm going to start looking into tweaking Intel CPUs too, just there isn't much to tweak with them really.

Also, I already know about CPU power measurement problems hence why I don't pay attention to them since too many of them are highly inaccurate with little repeatability, thus pointless to make a doodle about - but I found a problem with my own power measurement testing too. Unless people are uneducated about the whole testing procedures, there is no point paying attention to them until you can understand, repeat and verify them as holding any accuracy. The reviews/people doing power measurements and not double testing them with at least one other measurement method are going to end up with faulty readings too many times [too low or too high].
I checked CPU wattages with 3 CPUs with an AC power meter, a DC clamp meter and the Gigabyte P-Tuner wattage sensors with a (Q6600) at 1200MHz 0.925v. If 1.2v 2400MHz equals 95W TDP full load, then 0.925v 1200MHz would equal 28.2W TDP full load.
At the latter setting, idle, the power meter did not drop much AC Watts from CE1 stock idling. The clamp meter read 0.2A idling, which makes it around 2.4W DC idling. But when I checked with GBT P-Tuner being verified to read +12V1 and +12V2 very accurately [fault on +3.3V VDC), the values for DC Amps were 2.3A idling, which made it 27.6W DC, very close to CE1 stock power draw and verfiying exactly what my AC power meter read. Thus, unusual figures, especially low ones for non-low TDP CPUs are highly questionable.

Will be interesting to check out Phenom power draw at tweaked CnQ ~400MHz 0.6v, should outclass every major desktop model I can think of especially idling.

Undersea: http://www.xtremesystems.org/forums/...postcount=1594

[justapost]

Hmm canceled the online order and bought one in a shop.





CPU needs 1.475V for 3GHZ and get 75°C quickly under prime95 so i need better cooling now.

[aGeoM]

justapost Congrats man , yes your temps are high, humm, have you ever flashed a DFI bios in your board and compare the read temps? I ask this because your temps readings always higher than mines.
For the first impression, they are nice results, L3 6.69
I bought mine from ALTERNATE, only today HPM received them.

Good luck for your tests.

[Brother Esau]

Is anyone experiencing weird random issues with 3/31/08 Beta ?

[aGeoM]

Is anyone experiencing weird random issues with 3/31/08 Beta ?

Well, went back to 307. After a jumper clear CMOS, the system refused to boot @2750 (C1), flashed back 307 with /f command, and it's booting again @2750, in fact, just finished installing Server 2008 x64 at that config.

[Undersea]

9850 ordered! Should have it on Tuesday

We will see if the little brother board can do

[Oese]

Is anyone experiencing weird random issues with 3/31/08 Beta ?

no problems here until now..

[justapost]

justapost Congrats man , yes your temps are high, humm, have you ever flashed a DFI bios in your board and compare the read temps? I ask this because your temps readings always higher than mines.
For the first impression, they are nice results, L3 6.69
I bought mine from ALTERNATE, only today HPM received them.

Good luck for your tests.

Thx for the congrats. Nope did not try a Sapphire bios, my 9500 died too quickly with one of those. Gemini and BigTyphoon difference? Must check the comparison on anandtech.
I like HPM but i wanted one as quick as possible, fortunately a shop 15min away had a few.

Few results

CPU 2.8 NB 2.0 MEM 1066-5-5-5-16-30


CPU 2.8 NB 2.4 MEM 1066-5-5-5-16-30


CPU 2.8 NB 2.4 MEM 1066-4-5-5-16-30


So without further subtiming tweaking CAS 5 is faster than CAS 4 and 2.4 NB gives a good speed bump at least in crystal mark.

Wanted to bench 2-8/2.0/4-5-5-16-30 but i got an hd error so i'm in process of reinstalling.

@BroE; You have not apllied high voltages to your 9600BE yet, i'd recommend you sell that thing and get an 9850BE quick, much more fun and alot easier to oc.

3-31 is alot more unstable than 3-07 with high ref HT's here, but I gave it only a quickshot and did not try things like reflashing the bios.

[KTE]

^Agreed

Achim, check your 2G NB figures, the cache is ~7000 below mine at 2.5G CPU 2.0G NB 1066 5-5-5-15-11 Ganged. You should be plus 90k with that CPU speed.

[justapost]

^Agreed

Achim, check your 2G NB figures, the cache is ~7000 below mine at 2.5G CPU 2.0G NB 1066 5-5-5-15-11 Ganged. You should be plus 90k with that CPU speed.

Few more results:

5-5-5-16-30 RFC 105/75

MEM: 27795
CACHE: 91941

5-5-5-16-30 RFC 75/75

MEM: 29293
CACHE: 95681

5-5-5-16-11 RFC 75/75

MEM: 28462
CACHE: 82201

5-5-5-15-11 RFC 75/75

MEM: 31146
CACHE: 108764

[justapost]

Skipped crystalmark, it's too inaccurate for single runs.

With everest there is an improvement with 4-5-5-15 timings

Few results at 2.8GHz CPU 2.0/2.2/2.4GHz NB and 4-5-5-15-18 (left) / 5-5-5-15-18 (right) timings

CPU: 2.8GHz / NB: 2.0GHz


CPU: 2.8GHz / NB: 2.2GHz


CPU: 2.8GHz / NB: 2.4GHz


CPU: 2.9GHz / NB: 2.4GHz


CPU: 3.0GHz / NB: 2.4GHz


CPU: 2.793GHz / 2.128GHz NB / ref HT 266MHz / 533-4-4-4-12-18


CPU-Z Validation for 3.0GHz/2.4GHz


Seems the CPU needs more likely 1.5V to be stable at 3.0GHz now.

[Oese]

hehe, things cleared up a little over here.

After speaking with KTE i tested my dimms separately and they showed a real big difference..

one dimm is capable to run ddr1100 @ 2,07v just with max async set to 8ns, tras set to 16 and trc set to 13 everything else was left untouched so pretty friendly and understandable behaviour.

the other dimm is completely different and no way stable at ddr1100. so thats what kept me away from reaching desired ram clocks.

then i learned that obviously my pretty high vcore is required because of 4gb stressing the imc, with only one dimm the chip seems to be stable at 0,05v less. I'm still testing..

So finally i get to know this plattform really, never ran into such issues before on S939, but always theres a first time..

Thanks to KTE here and all of you guys!

[Brother Esau]

Yes Tye has been very instrumental in the Phenom/790FX situation since day one I tip my hat to you Tye and keep up the good work!

[Oese]

ok found the wall with my chip with only 2gb:

even at 1,6v no go...

but would be nice with a 2x1gb bench-set

but in a few days i'll test your 6400+ brother, its at the customs right now

[Brother Esau]

I just saw the thread updated with a post by you Oli and was going to ask if the CPU is in Germany yet. Good deal its a really nice CPU I know I really enjoyed it cool running and very fast

[Oese]

yepp i hope they dont mess with it at the customs and simply send it right to my door..

i really await it since due to the high multiplier i will be able to use my 4gb of ram with it at those high speeds man trust me I'm really anxious to test that chip in my system normally those high-end chips are theoretical experiences to me only thanks bro for that deal i just joined xs the right time methinks

[aGeoM]

Thx for the congrats. Nope did not try a Sapphire bios, my 9500 died too quickly with one of those. Gemini and BigTyphoon difference? Must check the comparison on anandtech.
I like HPM but i wanted one as quick as possible, fortunately a shop 15min away had a few.

Some reviews:

http://www.xbitlabs.com/articles/coo...geminii_6.html

http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

I use 2 fans, Silverstone FM121 + Enermax UC-12AEBS

Do you have the Sappire or the DFI? Coz R7LS307 it's Sappire bios.

THX

EDIT: BTW great work on 9850.

[justapost]

Some reviews:

http://www.xbitlabs.com/articles/coo...geminii_6.html

http://enthusiast.hardocp.com/articl...50aHVzaWFzdA==

I use 2 fans, Silverstone FM121 + Enermax UC-12AEBS

Thank you for the reviews. As I compared the fans I found that I picked the wrong ones I have SilenX Ixtrema Pro 74-14dB (72cfm) but should have taken the 76-18dB version (90cfm). Think I can expect a few C lesser temps with an FM121 (110cfm).
Thinkin about getting proper water cooling, but i'll wait for oldguys experiments with his vapo first.

Do you have the Sappire or the DFI? Coz R7LS307 it's Sappire bios.

Thx, fixed I had only changed the number going from Sapphire to DFI.

EDIT: BTW great work on 9850.

Hehe, thanks, that chips is alot smoother to oc. My 9600BE never liked more than ~1.42V, the 9850BE does not have this limitation.
I think this is the biggest improvement for ocers with the b3's.

@Oese, nice result with your 4600+, i wonder if the 6400 will get further.

[Oese]

@Oese, nice result with your 4600+, i wonder if the 6400 will get further.

I hope so... At least I will be able to run those speeds with 4gb of ram because i can avoid high htt because of the multiplier so i can expect not to run into imc problems, and it ran ~3,6ghz aircooled at be's system so thats what i hope for.. Perhaps a little higher?? And at lower vcore of course..

[Brother Esau]

My contribution towards the Sunday Gospel