** Official ** Gigabyte UD3 P45 Series -- EP45-UD3 / EP45-UD3R / EP45-UD3P

[NoGuru]

Thanks for the response.

I'm talking about the on board ehternet nic though. You ever use RJ45?

Some of the EP45's have problem with the nic's. Not sure how or if there is a fix other then RMA.

[zsamz_]

Thanks for the response.

I'm talking about the on board ehternet nic though. You ever use RJ45?

dissable then ennable in windows
i got same problem with evga p55 ftw
or you gotta rma

[In-Fluence]

Influence

This is the stock bandw i have at 4500 1600 CL 7-8-7-24 PL8 Turbo

some stability test at 24C ambient (thanx to my new system, could never do this before at this temp)


some everest bandw stable


this is not stable but nice bandw and 1T

Cheers
Sergio

Not bad at all, I did expect something a little higher but I'm guessing there's probably some bottlenecking when running DDR3 on P45. Thanks for the info once again

I noticed your 'write to precharge' and TWTr are a lot higher than DDR2, and they can affect bandwidth a lot, do they go lower at all?

All I gotta do now is wait for my good friend to come through for me.

[Dark-Energy]

Hey guys, I need your help on this. I picked up a Q9550 to replace my E8400 for my multi-threaded apps(converting to i7 = too much money cause of new board and memory), and along with that I purchased the ECO A.L.C. Liquid cooler (similar to the H50, which wasn't in stock). The batch number is L932B343 (packed September 2009) and it's an E0. When I checked the VID it was 1.150v, so I thought getting 500 FSB on my UD3P would be not all that hard(Isn't the UD3P one of the best for reaching high FSB frequencies on quads?).

Well, I cannot get it stable at 500 FSB no matter what settings I try(believe me, I have spent the last 10 hours on this with different settings and I'm about to pull my hair out). Even with the multi at 6x I could only get into windows by overvolting the VTT and MCH to 1.4v and then it still froze a second after prime 95 Blend. All of this instability results in freezing. I have only had a BSOD a few times, which was the pfn_list_corrupt and the other was the infamous uncorrectable hardware error.

I raised the vcore really high to 1.4 with LLC (around 1.38 in cpu-z) and got into windows at 4.0 GHZ and got an error after about 5 minutes in Prime Small FFT. But I don't understand this. It's a low VID chip. How come it needs 1.4 volts just for 4.0 GHZ? And Prime 95 blend was still freezing up too, so I'm at a loss. Could it just be my really old dusty memory? I've had it for over 2 years and I don't think it likes the 1:1 divider anymore especially at 500 FSB (although it worked with my E8400, but after extensive tweaking). It's 800mhz ram but it says it can run at 1000 with higher voltage and timings so maybe I've pushed it over its limits? But it's not failing memtest or anything so I have no idea what is wrong.

I have this weird gut feeling that this ram is somehow forcing my CPU to eat volts like crazy just to be semi-stable. I just don't want to believe that running at 500x8 should take that amount of voltage on such a low VID chip. Let alone, low vid chips are hotter, so my temps were pretty high when I did that.

Should I consider picking up a proper set of 1066 rated memory? Maybe it would solve half of my problems right there

[andressergio]

Not bad at all, I did expect something a little higher but I'm guessing there's probably some bottlenecking when running DDR3 on P45. Thanks for the info once again

I noticed your 'write to precharge' and TWTr are a lot higher than DDR2, and they can affect bandwidth a lot, do they go lower at all?

All I gotta do now is wait for my good friend to come through for me.

well yes mate me also but i didnt expect that P45 could run 1T will be nice on faster RAM

[NoGuru]

Hey guys, I need your help on this. I picked up a Q9550 to replace my E8400 for my multi-threaded apps(converting to i7 = too much money cause of new board and memory), and along with that I purchased the ECO A.L.C. Liquid cooler (similar to the H50, which wasn't in stock). The batch number is L932B343 (packed September 2009) and it's an E0. When I checked the VID it was 1.150v, so I thought getting 500 FSB on my UD3P would be not all that hard(Isn't the UD3P one of the best for reaching high FSB frequencies on quads?).

Well, I cannot get it stable at 500 FSB no matter what settings I try(believe me, I have spent the last 10 hours on this with different settings and I'm about to pull my hair out). Even with the multi at 6x I could only get into windows by overvolting the VTT and MCH to 1.4v and then it still froze a second after prime 95 Blend. All of this instability results in freezing. I have only had a BSOD a few times, which was the pfn_list_corrupt and the other was the infamous uncorrectable hardware error.

I raised the vcore really high to 1.4 with LLC (around 1.38 in cpu-z) and got into windows at 4.0 GHZ and got an error after about 5 minutes in Prime Small FFT. But I don't understand this. It's a low VID chip. How come it needs 1.4 volts just for 4.0 GHZ? And Prime 95 blend was still freezing up too, so I'm at a loss. Could it just be my really old dusty memory? I've had it for over 2 years and I don't think it likes the 1:1 divider anymore especially at 500 FSB (although it worked with my E8400, but after extensive tweaking). It's 800mhz ram but it says it can run at 1000 with higher voltage and timings so maybe I've pushed it over its limits? But it's not failing memtest or anything so I have no idea what is wrong.

I have this weird gut feeling that this ram is somehow forcing my CPU to eat volts like crazy just to be semi-stable. I just don't want to believe that running at 500x8 should take that amount of voltage on such a low VID chip. Let alone, low vid chips are hotter, so my temps were pretty high when I did that.

Should I consider picking up a proper set of 1066 rated memory? Maybe it would solve half of my problems right there

Well the E8400 is a very easy chip to overclock and does not take a lot of volts, were the Quads eat volt's for breakfast. It sounds to me like your barely giving it enough volt's. Tuning a quad takes very fine tuning of the CPU references as well.
Might be your ram too, but I would keep trying to get it stable before buying new ram. You should post you MIT and we can see where your settings are at.

[andressergio]

were the Quads eat volt's for breakfast

better than eat quads for breakfast

[radaja]

Influence

this is not stable but nice bandw and 1T




Cheers
Sergio

nice to see you can run 1twhen you get your new g.skill cl6's
it will be great at 2000 cl6 1t

[In-Fluence]

The best I managed with my extreme so far, the 4:5 divider needs quite a bit of MCH but vdimm is nice & low (and can probably go lower):


Yes 1T was always a nice boost way back when AMD ruled the roost and it always made the system much 'snappier'. Your new rams will no doubt get the same results, looking forward to seeing them!

[andressergio]

nice to see you can run 1twhen you get your new g.skill cl6's
it will be great at 2000 cl6 1t

Thanks Rob ! im happy with it let's see if the mobo can hold those beasts...

The best I managed with my extreme so far, the 4:5 divider needs quite a bit of MCH but vdimm is nice & low (and can probably go lower):


Yes 1T was always a nice boost way back when AMD ruled the roost and it always made the system much 'snappier'. Your new rams will no doubt get the same results, looking forward to seeing them!

Yeah i got more or less the same on DDR2 i started to love DDR3 of course on newer chipset perform faster but 11k and 46 lat will be awesome for P45

Cheers mates
Sergio

[OhNoes!]

Andres, I'm sure you can squeeze a little bit more out of it.

First image is EP45T-USB3P with G.Skill ECO Low Voltage @ DDR3-2000 CAS 7. Same RAM on EX58-UD3R @ DDR3-1600 CAS 6 in the second screnshot.

[andressergio]

Andres, I'm sure you can squeeze a little bit more out of it.

First image is EP45T-USB3P with G.Skill ECO Low Voltage @ DDR3-2000 CAS 7. Same RAM on EX58-UD3R @ DDR3-1600 CAS 6 in the second screnshot.

thanks mate i know but thats 8,5 multi...im at x9 wich is less bandw...

Thanks and nice bandw on the I7

Cheers
Sergio

[OhNoes!]

thanks mate i know but thats 8,5 multi...im at x9 wich is less bandw...

Thanks and nice bandw on the I7

Cheers
Sergio

Thanks. Yeah, they're both at 500FSB, and DDR3-2000. The difference is CAS 7 vs CAS 9. Your higher clock speed at CAS 7 should net you better read speeds, at least.

[andressergio]

Thanks. Yeah, they're both at 500FSB, and DDR3-2000. The difference is CAS 7 vs CAS 9. Your higher clock speed at CAS 7 should net you better read speeds, at least.

do you have the setting for 2000 CL7 stable ? i couldn do it yet on the USB3P

needs high volts on MCH and lots of vdimm or i am doing something wrong,,,also what Performace Level are u using ? if you can help me on a template or some settings will be awesome

Thanks mate
Regards
Sergio

[OhNoes!]

do you have the setting for 2000 CL7 stable ? i couldn do it yet on the USB3P

needs high volts on MCH and lots of vdimm or i am doing something wrong,,,also what Performace Level are u using ? if you can help me on a template or some settings will be awesome

Thanks mate
Regards
Sergio

My board is not up and running right now, but I did post my settings here: http://www.xtremesystems.org/forums/...postcount=6112

You may need to tweak a little bit more because of your higher clocks.

[andressergio]

My board is not up and running right now, but I did post my settings here: http://www.xtremesystems.org/forums/...postcount=6112

You may need to tweak a little bit more because of your higher clocks.

thanks but would be nice to know volts...

[NoGuru]

Serg will get it done, he is like Yoda!

[Dark-Energy]

Hey guys, I need your help on this. I picked up a Q9550 to replace my E8400 for my multi-threaded apps(converting to i7 = too much money cause of new board and memory), and along with that I purchased the ECO A.L.C. Liquid cooler (similar to the H50, which wasn't in stock). The batch number is L932B343 (packed September 2009) and it's an E0. When I checked the VID it was 1.150v, so I thought getting 500 FSB on my UD3P would be not all that hard(Isn't the UD3P one of the best for reaching high FSB frequencies on quads?).

Well, I cannot get it stable at 500 FSB no matter what settings I try(believe me, I have spent the last 10 hours on this with different settings and I'm about to pull my hair out). Even with the multi at 6x I could only get into windows by overvolting the VTT and MCH to 1.4v and then it still froze a second after prime 95 Blend. All of this instability results in freezing. I have only had a BSOD a few times, which was the pfn_list_corrupt and the other was the infamous uncorrectable hardware error.

I raised the vcore really high to 1.4 with LLC (around 1.38 in cpu-z) and got into windows at 4.0 GHZ and got an error after about 5 minutes in Prime Small FFT. But I don't understand this. It's a low VID chip. How come it needs 1.4 volts just for 4.0 GHZ? And Prime 95 blend was still freezing up too, so I'm at a loss. Could it just be my really old dusty memory? I've had it for over 2 years and I don't think it likes the 1:1 divider anymore especially at 500 FSB (although it worked with my E8400, but after extensive tweaking). It's 800mhz ram but it says it can run at 1000 with higher voltage and timings so maybe I've pushed it over its limits? But it's not failing memtest or anything so I have no idea what is wrong.

I have this weird gut feeling that this ram is somehow forcing my CPU to eat volts like crazy just to be semi-stable. I just don't want to believe that running at 500x8 should take that amount of voltage on such a low VID chip. Let alone, low vid chips are hotter, so my temps were pretty high when I did that.

Should I consider picking up a proper set of 1066 rated memory? Maybe it would solve half of my problems right there

Well the E8400 is a very easy chip to overclock and does not take a lot of volts, were the Quads eat volt's for breakfast. It sounds to me like your barely giving it enough volt's. Tuning a quad takes very fine tuning of the CPU references as well.
Might be your ram too, but I would keep trying to get it stable before buying new ram. You should post you MIT and we can see where your settings are at.

Update: So I decided just to bite the bullet anyways and went out and bought some new ram that was the Kingston KHX8500D2T1K2 1066mhz 2.2v

Anyways, I think it really helped in getting 500 FSB stable for once with reasonable Vcore. My first test was 450x8.5 which was a cakewalk and then 500x8 (4 GHZ) which required more Vcore than I thought just for a little bit more MHZ but I guess quads might need more volts the higher the FSB. Though it still doesn't explain why it needs just over 1.3 volts on a 1.15 VID chip, I've seen lower vcore at that FSB and multi on higher VID chips. Could somebody shed light on this please?

What got me prime blend stable and Linpack stable was screwing around with the clock skews. First I set MCH clock skew to 50ps; blend went from freezing right away to freezing after about 8 minutes. I then upped it to 100ps and added 50ps to the CPU clock skew and I ran blend for 30 minutes and it passed, same with 5 passes of Linpack. Still doesn't mean it's completely stable but I guess I'm at a good starting point.

Here's my settings by the way. I tried to remember them as best as I could.

******Motherboard Intelligent Tweaker (M.I.T.)******
 
Robust Graphics Booster.........................[ Auto     ]
CPU Clock Ratio.................................[ 8x       ]
Fine CPU Clock Ratio............................[ + 0.0    ]
CPU Frequency 4.0GHz............................[ 500 x 8.0  ]
 
******Clock Chip Control******
 
CPU Host Clock Control..........................[ Enabled  ]
CPU Host Frequency (Mhz)........................[ 500   ]
PCI Express Frequency (Mhz).....................[ Auto  ] (Not sure if setting to 100 or 101 would help)
C.I.A 2 [Disabled] 
 
******Advanced Clock Control******

CPU Clock Drive.................................[ 800mv] (was at 700mv for some reason and couldn't even get past the loading screen. Though trying 900 gave the same results)
PCI Express Clock Drive.........................[ 900mv]
CPU Clock Skew..................................[ 50ps]
MCH Clock Skew..................................[ 100ps]
 
******DRAM Performance Control******
 
Performance Enhance.............................[ Standard/Turbo (Can't remember) ]
Extreme Memory Profile (XMP)....................[ Auto  ]
(G) MCH Frequency Latch.........................[ Auto   ]
System Memory Multiplier........................[ 2.00B ]
Memory Frequency   ..............................[ 1000 ]
DRAM Timing Selectable..........................[ Manual  ]
 
CAS Latency Time................[ 5    ]........[ 5     ]
tRCD...........................[ 5    ].........[ 5     ]
tRP.............................[ 5    ].........[ 5     ]
tRAS...........................[ 15   ]........[ 15     ]
tRFC...........................[ 52   ].........[ auto   ] (says 64 in CPU-z)
Command Rate (cmd).............[ 0   ]..........[ auto  ] (says 2T in CPU-z)
 
******Motherboard Voltage Control******
 
Load-Line Calibration...........................[ Enabled  ]
CPU Vcore         [ 1.25    ]...................[ 1.35625]
CPU Termination   [ 1.20    ]...................[ 1.34    ]
CPU PLL           [ 1.50    ]...................[ 1.5    ]
CPU Reference     [ 0.760   ]...................[ .785 at 1.2 Vtt     ] (haven't really experimented with this much but I heard this was a good starting point)
 
MCH Core          [ 1.100   ]...................[ 1.36   ]
MCH Reference     [ 0.760   ]...................[ .800 at 1.2 Vtt   ] (any lower causes erroring in Prime 95 Blend
MCH/DRAM Refernce [ 0.900   ]...................[ auto    ]
ICH I/O           [ 1.500   ]...................[ auto   ]
ICH Core          [ 1.100   ]...................[ auto   ]

DRAM Voltage      [ 1.8    ]...................[ 2.2  ] 
DRAM Termination  [     ]...................[ auto  ]
Channel A Ref     [     ]...................[ auto  ]
Channel B Ref     [     ]...................[ auto  ]

*****************Advanced BIOS Features********* 

Limit CPUID Max to 3 ................. [disabled] 
No-Execute memory Protect ............ [disabled] 
CPU Enhance Halt (CIE) ............... [disabled] 
C2/C2E State Support ................. [disabled] 
C4/C4E State Support ................. [disabled] 
CPU Thermal Monitor................... [disabled] 
CPU EIST Function..................... [disabled]
Virtualization Technology ............ [disabled]

LinX brings my temps up to a max of 66/66/68/68 is this okay? I want to push to 4250 eventually but I'm not sure if I can get that now. I wasn't expecting it to get this hot especially at only 1.328 volts but maybe the low VID at 1.150v has something to do with it, which is why I am certain there's a way to get this below 1.3v. I'm on integrated liquid cooling too, with 2 fans on either side of the rad pulling in from the outside and into the case and out the top so I was expecting a bit better temps.

[In-Fluence]

Hi mate. Temps are great considering it's linx, The low 80's is the absolute limit whilst running this imo so you've got plenty of headroom. Some say low vids are often a sign that it will run high clocks on low volts, and/or can also get hotter than higher vid chips. I'm sure you can push further and you've got it at a nice, low voltage for that speed, so go for more! Some DDR1200 would have been a better combo, but it seems your chip is set to do well, and you could look at pushing quite a lot further with some careful tweaking

[NoGuru]

Dark Energy, read this and see if it helps any http://www.overclock.net/intel-mothe...-ep45-ud3.html

Here is a pic of my rig now, before all my WC parts come it.

[andressergio]

Dark Energy, read this and see if it helps any http://www.overclock.net/intel-mothe...-ep45-ud3.html

Here is a pic of my rig now, before all my WC parts come it.

grats my friend will look awesome watercooled

[Dark-Energy]

Hi mate. Temps are great considering it's linx, The low 80's is the absolute limit whilst running this imo so you've got plenty of headroom. Some say low vids are often a sign that it will run high clocks on low volts, and/or can also get hotter than higher vid chips. I'm sure you can push further and you've got it at a nice, low voltage for that speed, so go for more! Some DDR1200 would have been a better combo, but it seems your chip is set to do well, and you could look at pushing quite a lot further with some careful tweaking

Well I looked and they didn't have any DRR2-1200 ram so I just went with the 1066 but I could even push to 540 FSB if I wanted to quickly bench because I found the ram is stable at 1080. But I doubt anything above 520 FSB will ever be completely stable for the CPU unless I get lucky and spend days of tweaking

Dark Energy, read this and see if it helps any http://www.overclock.net/intel-mothe...-ep45-ud3.html

Thanks

[andressergio]

Sup guys for those on the EP45T USB3P i updated to F2D bios, no issues at all

got a nice run at 24C last nite on low vcore



will go up till 4608 lets see how it goes

Cheers
Sergio

[andressergio]

been testing a while this mobo and i love it , its like a remake of the good UD3P rev 1.0

been testing all that i did on the UD3P rev 1.0 and no problems at all, i remember testing the limits on in and goot some bad reboot loops, this doesn't have that problems, sometimes when pushed hard yes you just power off and back, but had to do that only 2 times in 2 weeks of testing rest it autorecovers pretty fast...

here's what i am testing now, mem is not tweaked just using auto and turbo



some bandwith




Cheers
Sergio

[OC Nub]

Nice job Sergio. Look like a fun board to play with.