|
Post by somerandomtechyboi on Feb 12, 2022 7:38:54 GMT -5
Done a bit more testing and max stable i got is 556fsb, its ok i guess but i was hoping for higher, max i got that could actually boot into the bios is 610fsb
Also why does my cpu require less volt to run 4.4ghz on this board than the p5q? Cause on the p5q it needed ~1.43v but on this ds4p it only needs ~1.37v
|
|
|
Post by Vinster on Feb 12, 2022 21:29:07 GMT -5
different board have different power delivery systems. some boards can handle the vdroop better than others... happens when big loads shift better than others.
Vin
|
|
|
Post by somerandomtechyboi on Mar 7, 2022 1:35:55 GMT -5
Now i got a p6t deluxe v2 (ofc flashed to p6x58d premium immedeatly) and after booting 5g the first hour i got it and screwing around with efficiency oc, got bored and now ive stuck the bios chip in my p5q and flashed a p5q deluxe bios (ebb changed to maximus ii formula), it works but stupid thing keeps booting into ez flash constantly, just a minor inconvenience tho and it still works for oc, same fsb oc unfortunately but now i have access to the extra settings like gtls and clock skews so maybe itll help once i tec cool the nb. Im currently just plugging all the settings from my ds4p onto this thing as a reference, my ds4p usually needs 0.51x nb gtl but the p5q needs 0.52x, do i need to offset the gtls by +0.01? Cause seems like it needs higher gtl compared to my ds4p and playing with gtls hasnt gotten any fsb improvement yet
|
|
|
Post by Vinster on Mar 7, 2022 8:12:38 GMT -5
do i need to offset the gtls by +0.01? Cause seems like it needs higher gtl compared to my ds4p and playing with gtls hasnt gotten any fsb improvement yet Every board is different. I'd never expect settings from one board to work verbatim on another. It's only a starting point.
Vin
|
|
|
Post by somerandomtechyboi on Mar 7, 2022 9:35:11 GMT -5
Yea the p5q deluxe bios doesnt seem to be helping at all, even with the gtl settings, too bad for whatever reason clock over charging volt is not adjustable by the bios, oh well gotta love crappy asus bios.-.
|
|
|
Post by Vinster on Mar 7, 2022 13:26:44 GMT -5
not always the bios... some boards just don't like gtl settings... I have a Gigabyte UD3L that doesn't use any GTL setting I toss at it..
Vin
|
|
|
Post by Macsbeach98 on Mar 7, 2022 19:47:19 GMT -5
P45 DDR2 sucks I just couldnt put nitrogen on that P5Q its a waste of time same as UD3P when I can put nitrogen on X48 go to 640 bus with 2 sticks of ram in and run benches reliably and stably and dont touch a GTL setting at all. EP45T-UD3R the DDR3 boards are a lot better than the DDR2 boards but still get their asses handed to them by X48 when running benchmarks.
I just bought a P5E Deluxe X48 DDR2 it has the same heatsink asssembly as my Rampage Formula I hope it goes as good.
|
|
|
Post by somerandomtechyboi on Mar 7, 2022 21:13:50 GMT -5
P45 DDR2 sucks I just couldnt put nitrogen on that P5Q its a waste of time same as UD3P when I can put nitrogen on X48 go to 640 bus with 2 sticks of ram in and run benches reliably and stably and dont touch a GTL setting at all. EP45T-UD3R the DDR3 boards are a lot better than the DDR2 boards but still get their asses handed to them by X48 when running benchmarks. I just bought a P5E Deluxe X48 DDR2 it has the same heatsink asssembly as my Rampage Formula I hope it goes as good. I think you can flash the bios straight to a rampage formula/extreme with a bios programmer
|
|
|
Post by Vinster on Mar 7, 2022 23:53:15 GMT -5
I had a P5E Deluxe.. the MB would not go over 400mhz FSB... on any CPU.. I didn't know you could cross flash it... I don't have that board anymore, but I wonder if that would had sorted it out... hmm... oh well..
Vin
|
|
|
Post by Bones on Mar 8, 2022 0:24:36 GMT -5
P45 DDR2 sucks I just couldnt put nitrogen on that P5Q its a waste of time same as UD3P when I can put nitrogen on X48 go to 640 bus with 2 sticks of ram in and run benches reliably and stably and dont touch a GTL setting at all. EP45T-UD3R the DDR3 boards are a lot better than the DDR2 boards but still get their asses handed to them by X48 when running benchmarks. I just bought a P5E Deluxe X48 DDR2 it has the same heatsink asssembly as my Rampage Formula I hope it goes as good. I think you can flash the bios straight to a rampage formula/extreme with a bios programmer I don't believe you can flash a REX BIOS to a Formula board due to the Formula being a DDR2 board, the REX itself being a DDR3 board. I mean you CAN flash the BIOS chip itself but it probrably won't work when you try to start the system. What you have pictured could work though.
|
|
|
Post by somerandomtechyboi on Mar 8, 2022 1:04:59 GMT -5
I think you can flash the bios straight to a rampage formula/extreme with a bios programmer I don't believe you can flash a REX BIOS to a Formula board due to the Formula being a DDR2 board, the REX itself being a DDR3 board. I mean you CAN flash the BIOS chip itself but it probrably won't work when you try to start the system. What you have pictured could work though. Whoops Forgot about the REX being a ddr3 board, but u guess the formula could work as you said, also i cant run a maximus ii bios cause it just gives 0E code so i habe to stick with the p5q deluxe bios which doesnt seem to be helping at all with ocing
|
|
|
Post by Macsbeach98 on Mar 8, 2022 2:08:13 GMT -5
Once I have got the board I will sit it next to the Formula and see if everything is the same then I will think about flashing the formula bios to it.
|
|
|
Post by osmiumoc on Mar 8, 2022 9:51:32 GMT -5
P45 DDR2 sucks I just couldnt put nitrogen on that P5Q its a waste of time same as UD3P when I can put nitrogen on X48 go to 640 bus with 2 sticks of ram in and run benches reliably and stably and dont touch a GTL setting at all. EP45T-UD3R the DDR3 boards are a lot better than the DDR2 boards but still get their asses handed to them by X48 when running benchmarks. I just bought a P5E Deluxe X48 DDR2 it has the same heatsink asssembly as my Rampage Formula I hope it goes as good. Performance wise X48 is king. And any DDR3 setups beats DDR2 as well. However, somehow P45 seems to be the best if you want to go for high DDR2 clocks. All 3 top frequency scores were made on P45 boards. Two of them in dual channel. 700MHz (DDR2 1400) CL4 was achieved with P45. In the top 16 FSB frequency records you can find only 1 X48 result at rank 12. Every other board is P45. Like I said in my post on the EP45-Extreme thread, there has to be a way to make them work. Either I have to try magic and study forbidden arcane knowledge or its just a matter of throwing thousands of hours trial and error at this platform.
|
|
|
Post by PolRoger on Mar 8, 2022 11:19:47 GMT -5
Once I have got the board I will sit it next to the Formula and see if everything is the same then I will think about flashing the formula bios to it. I picked up an ASUS P5E Deluxe combo off of eBay back in October and I had been wanting to flash it to a Rampage BIOS. The Deluxe and the Formula boards are very similar the Formula has Start/Reset buttons while the Deluxe has bare silk screen indicators on the board. The Formula also has 3 optional temp sensors that the Deluxe doesn't have. There is also two bare silk screen indicators on the Deluxe board near the edge close to the first PCIE Supreme FX sound card slot. I followed Tagg's Video and Flashed it over today. Everything is looking good so far. When I first booted the board up after the Flash the BIOS was showing a high OPT 1 Temp Alert? So I just disabled that sensor in BIOS.
|
|
|
Post by somerandomtechyboi on Mar 8, 2022 21:07:38 GMT -5
P45 DDR2 sucks I just couldnt put nitrogen on that P5Q its a waste of time same as UD3P when I can put nitrogen on X48 go to 640 bus with 2 sticks of ram in and run benches reliably and stably and dont touch a GTL setting at all. EP45T-UD3R the DDR3 boards are a lot better than the DDR2 boards but still get their asses handed to them by X48 when running benchmarks. I just bought a P5E Deluxe X48 DDR2 it has the same heatsink asssembly as my Rampage Formula I hope it goes as good. Performance wise X48 is king. And any DDR3 setups beats DDR2 as well. However, somehow P45 seems to be the best if you want to go for high DDR2 clocks. All 3 top frequency scores were made on P45 boards. Two of them in dual channel. 700MHz (DDR2 1400) CL4 was achieved with P45. In the top 16 FSB frequency records you can find only 1 X48 result at rank 12. Every other board is P45. Like I said in my post on the EP45-Extreme thread, there has to be a way to make them work. Either I have to try magic and study forbidden arcane knowledge or its just a matter of throwing thousands of hours trial and error at this platform. Weeps in ~1500 7-9-9 extreem dark Im just trying to clock the balls off these rams and see if i can get 1500+ stable, havent been succesful yet cause it just keeps freezing i assume these are nanya d die but they are kinda meh with timings, speed is ok but timings suck
|
|
|
Post by somerandomtechyboi on Mar 13, 2022 2:50:29 GMT -5
Ok seems like the p5qd bios is actually better than the deluxe for oc surprisingly, prob just me using the deluxe bios wrong cause didnt screw around with clock skew but im too lazy for that not to mention what the hell am i even aiming for on the ram clock skews
Now i decided to just max out all the secondaries and it seems to be helping in ram oc, though what are the secondary timings that affect ram clocking capability the most? Cause it seems like by yeeting the secondaries i have a boatload of headroom, pretty sure i can do 1600mhz now, maybe even try stabilizing it
|
|
|
Post by somerandomtechyboi on Mar 19, 2022 11:11:11 GMT -5
I think i forgot to tell you guys that i now have a p6t deluxe + x5660, some of you prob already know as i also post in ltt forum and even the hwbot forum but basically just rip cmos, rip corner, rip half the i/o, rip main pcie slot, it still works, booted to 5g in 1 hour with ~1.5v vcore (crashed immedeatly after), then started testing only to find out uncore is rubbish (4g stable at 1.5v imc/qpi) and somehow run my samsung 1gbit rev f to 2200mhz 9-11-11-24 stable at 2v (prime95 large ffts) but caps at 2200 no matter the timing, not bad for a generic stick but could be better
Now im wondering on the diff between xeon l, e, x,and w series, im mainly looking for better uncore overclocking ability and maybe better core overclocking ability (this cpu is actually somewhat decent as long as i keep temps low, 4.3g ~1.3v is my estimation on 6 cores under 80c but atm cause sht cooler 3 cores 4.3g at 1.288v 85c hottest core).
The e xeons in 775 for all i know were considered to be good bin cause lower tdp than the x cpus which are pretty much the same thing but higher tdp (like x5450 vs e5450)
The l xeons seem to heat up like crazy at high clockspeeds but unsure if thats the case for all l xeons
W xeons have more ram dividers and unlocked multi but idk if there are any other diffs between w and x
|
|
|
Post by Mr.Scott on Mar 19, 2022 11:33:41 GMT -5
OC ability and uncore is luck of the draw. They all run hot. 4.3 24/7 on air is about the norm.
|
|
|
Post by somerandomtechyboi on Mar 21, 2022 11:13:45 GMT -5
OC ability and uncore is luck of the draw. They all run hot. 4.3 24/7 on air is about the norm. Well it does run 4.5 at 1.38v on its 2 best cores and from the 4.3g results adding the 3rd core to give more accurate oc estimations, on the 2 best cores 4.3g requires 1.256v but with the mediocre 3rd core added it jumps to 1.288v so with that id assume 4.5g would be ~1.42v with 3 cores and ~1.45v with all 6 cores or if the worst core turns out to be that bad then 5 cores Would ~1.45v for 4.5g be average or slightly above average for a xeon? Cause atm im just abusing the sht out of it with 1.55v qpi and 4000mhz uncors + 2200mhz generic ram shenanigans
|
|
|
Post by Mr.Scott on Mar 21, 2022 12:42:43 GMT -5
To be above average you would need to be 4.5 @ 1.35v
Just for example, I am at 4.3 @ 1.3v on air and I consider that average.
|
|
|
Post by somerandomtechyboi on Mar 21, 2022 13:22:31 GMT -5
To be above average you would need to be 4.5 @ 1.35v Just for example, I am at 4.3 @ 1.3v on air and I consider that average. Welp my cpu for some reason keeps crashing on core 3 even at 4.2g 1.3v so if i cant fix this ill just regard this x5660 as a pos and swap it for something better or attempt to get lucky on another cpu like an e5649 or x5650
|
|
|
Post by somerandomtechyboi on Mar 24, 2022 19:08:43 GMT -5
Ok now i have another question on x58 xeons Any diff in the 4 core xeons binning wise compared to the 6 core xeons? Highest bin 4 core i have access to is the xeon e5640
Also how good or bad does ecc unbuffered ddr3 overclock? Cause they are cheap but so far ive only seen them with samsung 1gbit rev e, could these unbuffered sticks come with good dies like 1gbit/2gbit samsung rev d?
|
|
|
Post by Vinster on Mar 24, 2022 20:46:27 GMT -5
Ok now i have another question on x58 xeons Any diff in the 4 core xeons binning wise compared to the 6 core xeons? Highest bin 4 core i have access to is the xeon e5640 Also how good or bad does ecc unbuffered ddr3 overclock? Cause they are cheap but so far ive only seen them with samsung 1gbit rev e, could these unbuffered sticks come with good dies like 1gbit/2gbit samsung rev d? The tinkering I've done, Xeons are their own beasts to a point compared to a commercial counterpart and shouldn't be compared to them..
with Xeons, the memory multipliers are locked, so stick your best (fastest) memory with them as you might have to clock them high.
Lastly, because or the error checking in ECC ram, I wouldn't bother trying to OC it... Even if you did get it to fast speeds, the latency would be slower because of the checking and would effectively be slower.
Vin
|
|
|
Post by somerandomtechyboi on Mar 25, 2022 2:12:36 GMT -5
Ok now i have another question on x58 xeons Any diff in the 4 core xeons binning wise compared to the 6 core xeons? Highest bin 4 core i have access to is the xeon e5640 Also how good or bad does ecc unbuffered ddr3 overclock? Cause they are cheap but so far ive only seen them with samsung 1gbit rev e, could these unbuffered sticks come with good dies like 1gbit/2gbit samsung rev d? The tinkering I've done, Xeons are their own beasts to a point compared to a commercial counterpart and shouldn't be compared to them..
with Xeons, the memory multipliers are locked, so stick your best (fastest) memory with them as you might have to clock them high.
Lastly, because or the error checking in ECC ram, I wouldn't bother trying to OC it... Even if you did get it to fast speeds, the latency would be slower because of the checking and would effectively be slower.
Vin
Though im not talking registered ecc, just unbuffered ecc with the extra ram ic so one side has 5 ics while the other has 4 ics, though would unbuffered ecc help with preventing crashes or something? Cause it would be pretty useless if it didnt help prevent random crashing but then again i can get em for dirt cheap Now im also considering elpida hyper rams, though since ill be running pretty high bclk (200-224bclk) what could i expect out of them? Cause any less than 2200 will start limiting my bclk oc cause atm i can run my generic rev f to 2200 with 9-11-11-24 trfc 160 at 2v but no matter what ive tried it would not go over 2200mhz and im already at the voltage ceiling and can only increase by 0.04v since i ran this speed at 220bclk w 4000mhz uncore. Also is 3000mhz ddr3 possible on x58? Cause these imcs are pretty rubbish for all i know, if i could run 3000mhz then maybe id consider 2gbit samsung rev d
|
|
|
Post by Macsbeach98 on Mar 25, 2022 2:35:27 GMT -5
All ECC ram does error checking. What you need are Hypers or PSC. Hypers can run really tight timings up to around 2200 PSC can run upto around 2600. I am running PSC with 4790k in Challenger Div V 8-11-8-24-1T at 2400mhz the CPU is old and has been frozen umteen times used to run 2600mhz 8-12-8-28. Oh and I am winning. 3000mhz ram on X58 yea right with one stick you will probably not get much more than 2200mhz Vin W series Xeons are unlocked I was running 197 base 31x multi 6100mhz for pifast in Shrimps comp.
|
|
|
Post by Vinster on Mar 25, 2022 9:56:20 GMT -5
Vin W series Xeons are unlocked I was running 197 base 31x multi 6100mhz for pifast in Shrimps comp. Yes, that's true. I was thinking of Server Xeons that are the X or L series.
Vin
|
|
|
Post by Vinster on Mar 25, 2022 10:00:24 GMT -5
on x58, once you reach 4000mhz uncore, it's very cpu bin dependent on how much higher you can get.
Typically, once you get to 4000mhz, the voltage and cooling required to go higher is a mountain of effort with little to no performance gain. it's an X58 limitation.
you're best off then decreasing your uncore multiplier to keep clocking up.
Vin
|
|
|
Post by Mr.Scott on Mar 25, 2022 19:19:38 GMT -5
I find that anything much over 3800 uncore is worthless anyway. At least in benchmarks.
|
|
|
Post by Vinster on Mar 25, 2022 21:01:06 GMT -5
I find that anything much over 3800 uncore is worthless anyway. At least in benchmarks. Thinking more about it. the gains I recall from memory were within error (1-2%)
thing that sucked was being near 4000mhz and ticking the uncore multi down one tick was a big difference that was hard if not impossible to recover from. I was always at my thermal capability and there was no way to re-gain the loss...
I usually went back to 12x or 14x (which ever I was on), went as close as I could to 4000 and left it at that.
Vin
|
|
|
Post by somerandomtechyboi on Mar 25, 2022 23:35:12 GMT -5
I guess my cpu is either an average sample or below average sample then, welp guess i have nothing to lose by abusing it anyways so im free to experiment without fear of destroying a good bin cpu, maybe ill try selling this cpu and swapping it for an e5649 and see if i can get a better bin cpu
Also why the hell does the bios not let me set 22 multi? It jumps from 21 to 23 multi which is not helpful, ive even seen it go to 24 multi but for whatever reason i usually also cant set 24 multi
|
|