@JoBigShow
Power Requirement for 4870 CF
http://www.amd.com/us/products/desktop/gra...quirements.aspx

i checked the FSP manual. I believe your PSU's +12v rail is capable handling more than 35A (the 35A is the minimum and the maximum will be around 47A~48A). See at +5v,+3.3v, +12v combined wattage is 575W and minimum +5v & 3.3v(in the manual) is 5W and maximum 155W (on the PSU label) but your PSU label is quite confusing because it doesn't state the maximum combined +12v rail like other PSUs ie Gigabyte PSU we discussed earlier... its +12v rail has 30A max. For safety... 35A is the best assumption...
Your PSU absolutely can handle 4870 CF and still got a few headroom for other hardwares ie HDD... so don't worry...

This post has been edited by campsol2k: Sep 7 2009, 12:46 PM

I'm relief that the PSU can handle 2 of them GC's. I thought i bought the wrong PSU bcoz i always hear people rave about that u need minimum 750w PSU for SLI and CF. (Ok maybe if they mean like SLI'ed GTX285 or Tri SLI'ed GTX260 than thats understandable...)

How did u calculate the max amps of +12v rail? The only info i get is from the manual book and is on the reference of Over Current Protection, which stated that:
OUTPUT VOLTAGE
+12v1,2,3, & 4 = 30A each max. current limit
+3.3v = 60A max. current limit
+5v = 48A max. current limit
Even in the manual it doesnt state about max Amps on +12v rails, so i deduce that its couple of Amps below the when the OCP kicks in.

Ok. Here's the tricky part when selecting a PSU....
some of the reason why ppl recommend such huge amount of wattage because of the amount of native PCI-E connectors that came together when u purchase a PSU.
ie.. when u want to CF 4870... u need 4X 6-pin PCI-E connector... but most PSUs below 700~750W dont provide sufficient connectors...
i don't have any definitive/exact reasons why PSU manufacturers dont provide enough amount of connectors although the total wattage capacity is sufficient to power two ATI HD4870 cards in CF mode... but i believe the answer lies at the PSU design itself... i'm not a PSU engineer or designer. So i only can make my own assumption based on my readings.. and i believe the reason why they do this is because of the railing/design issue ...

My calculation was based on the "minimum load" column under electrical specs table. Don't take for granted what i said earlier... i calculated the max wattage based on my assumption. That's why i can come up with my own assumption of the current range(35A~48A)... Based on my research that i've done... most 600W PSU usually can handle around 40A@12V rail. Take note that... a 600W PSU isnt made the same as other 600W PSU... that's why u saw variety of 600W PSU.. some have multiple rail +12V .. some have single rail +12V... some support large current@+12V rail... some arent...

i forgot to tell u that when u CF 4870 using your PSU.. the 12V rail will be heavily loaded...
ie..
i only take the significant power consuming hardware only
CPU - 40W
2X GC - 360W
Total - 400W max... 20W headroom if we assume your PSU can handle 35A@12V rail...

I hope we don't stray out of topic too long or else the moderator will notice us... it should be inside this thread http://forum.lowyat.net/topic/450882 or http://forum.lowyat.net/topic/376014

This post has been edited by campsol2k: Sep 8 2009, 06:26 AM

Alright bro, I'll migrate to the threads you showed me, the moderators wont be to pleased for strayed topics.
Anyways I read reviews that the power consumption of 4870CF are around 270w~300w, for example:

I think my PSU is sufficient to handle them to GC's. I turst your judgments bro.
I will venture out into International forums for more info on the FPS Epsilon 600..
Thanqs for all the abundance of help you have given bro





I'm back for E7xxx OC discussions . Will post my latest development of E7400 OC results soon...

Alright bro, I'll migrate to the threads you showed me, the moderators wont be to pleased for strayed topics.
Anyways I read reviews that the power consumption of 4870CF are around 270w~300w, for example:

I think my PSU is sufficient to handle them to GC's. I turst your judgments bro.
I will venture out into International forums for more info on the FPS Epsilon 600..
Thanqs for all the abundance of help you have given bro





I'm back for E7xxx OC discussions . Will post my new E7400 results soon...

---

Added on September 8, 2009, 8:15 am
Sorry for Double post guys, my Internet connection is sluggish these days...

This post has been edited by JoBigShow: Sep 8 2009, 08:15 AM

OC TRIAL #3

FSB 400 x 9.5 = 3.8ghz @ 1.296v (BIOS)
Run Time on Linx: 28m 34s

Main Components in Rig:
E7400 3.8Ghz @ 1.280v, with AC Freezer Xtreme
Asus P5QL-PRO
2G Kingston HyperX 800mhz
FSP Epsilon 80Plus 600W



Report:
I have decrease the Vcore to the point of minimum voltages that the E7400 can handle @ 3.8Ghz:
1.280v. in CPU-Z; 1.296v in BIOS, which is about just one increment away compared with my previous 'OC TRIAL #2' @ 1.3v BIOS

The Vcore hovers around 1.28v to 1.292v in CPU-Z during the 20 rounds of LinX test runs.
I'm really stocked with the maximum temperature readings by only reaching @72c this time,
probably due to the lower voltage than my previously OC TRIALs, and not to forgot, the good work of the AC Freezer Xtreme.
On average the temps are approximately around 60c~65c on the LinX test runs.

I will humbly accept this OC setup because this is my main rig and I intend to run it 6hours+ average, per day.
Therefore I will stick to this setup, for now.

-------

Talking about 4Ghz as benchmarking, I still could not ascertain any stability within those Ghz range. My hypothesis is that due to motherboard lacking the capability to OC 400 X 10. So in the next benchmarking test, I will try varieties of FSB X multiplier combinations to attain the 4Ghz mark, or at least the closest as possible.
I probably need a P45 chipset motherboard to OC seriously. But then again like I mentioned, I'm fine with 3.8ghz system for daily usage........



This post has been edited by JoBigShow: Sep 9 2009, 08:57 AM

Tight the ram timing bro since you already decided to stick w this profile.
Your ram latency is high.

-pWs-

Yes, i've been working on it these days, but got into some setbacks, I've posted i thread at Q & A on how much voltage i need to stabilize the RAM from 5-5-5-15 to 4-4-4-12:
http://forum.lowyat.net/index.php?act=ST&f=116&t=1157344

At the moment, i'm trying to change my current 5-5-5-'18' to 5-5-5-'15', but strangely I already set the 'CYCLE TIME (tRAS)' to 15 in BIOS and it still shown 5-5-5-'18' in CPU-Z...
Can you tell my what is up bro??

nice job there..



-pWs- do u recommend to OC with LLC enabled or disabled? currently i m on 333x10.5 with LLC disabled.. with a Vdroop about 0.03V..

can the timing of KVR be tighten as well?

Nobody can tell you how much vDimm needed. You have to test yourself.
For the tRas, it could be Bios issue.

http://www.overclock.net/intel-motherboard...s-llc-help.html
Here you go.

-pWs-

Ok bro, I'm starting @ 2.2v as a 'control', and will tune it accordingly.
When u said nobody can tell, do you mean each person's PC is different in terms of their systems?
hence different systems have different demand in vDimm?, which then ultimately lead me to believe 'Reviews' of products(especially on DRAM) arent to be trusted 100%?
Probably they shouldn't be trusted 100% in the first place, but I need at least somewhere to begin with...so thanks bro for telling me different RAM have different vDimm requirements. I thought only processors are unique to one another.
With this somewhat seems insignificant little info is making me progress a lot atm.
Now my OC Rig is somewhat near completion.....for now.

Here my updated 4-4-412 settings @ 2.2v


I will Run ORTHOS or prime95 to obtain the result of its stability.
(currently had only been running on prime95 for just 30minutes)

I will post the results soon

tq tq..

juz wanna noe, do u enable it?

Every ram different vDimm based on the chipset build in on the ram.
If same chipset, it might use the more o less same vDimm. [For reference]

Different platform yields different vDimm also. Bear in mind.
A lot of aspects affects too. Ram timng, ram speed or both of them affect most.

Slowly find out by using MemTest. Make sure you fully test the unused ram.

-pWs-

I din come across this setting in my board. Or maybe I miss it.

-pWs-

oh ok.. but some board use other terms if not mistaken.. if no LLC function means it is same as disabled?

wat's yr comment on my Vdroop?

to get 333x10.5 stable,

my setting in bios is 1.28750V, cpu-z shows 1.252V max 1.216V min.. LLC disabled..

On Quad?
If on dual, it is a lot.

-pWs-

obviously on dual since we r in E7XXX thread..

mine is E7400 anyway.. yea, kinda high too IMO

before this i'm still able to get 300x10.5 at stock Vcore, then when i hit 333, i was shock i need 1.2875V to get it stable during stress test..

anything lower than that will not get me restart, but always failed in LinX test after 10 loops or less..

---

Added on September 9, 2009, 12:42 pmforget to mention with LLC disabled, I believe, IF i enable LLC, i will be able to get a lower stable Vcore..

This post has been edited by farscope: Sep 9 2009, 12:42 PM

Ops. Sorry. Never notice we are in E74.
If then, try to enable LLC and see got improve the vDroop or not.

-pWs-

will continue OC and test out the LLC enable effect when i feel 333x10.5 is not enuf..

my next OC will try to hit 400x9 or more to get 1:1..

as for now, i'm done with 333x10.5.. will update bout the LLC enabled function with lower Vcore and see how it goes..

Just in case if want to know, i Enabled mine, since LLC (Load Line Calibration) claims to improve V-drop directly, (its what the info written in my bios)
When I enabled it, my V-drop 'drop' around from 1.3v(bios) to 1.296v (CPU-Z). Have you ever try 'enable' that setting bro?