TPU's Nostalgic Hardware Club

[MachineLearning]

CPU-Z Validation

Attached Cinebench R23.2, CPU-Z benchmark, and a couple pictures of the build... aand it's probably my least favorite I've ever done, I really dislike this case and its fans. Not fun to work in, not quality and the whole thing has no aesthetic, but that's kinda charming too I suppose. I'm not sure if this is a great sample or not, as I'm not very into AMD CPU overclocking yet.

Though no matter what speed they're at, FX CPUs play like it's golf...

Full specs:
- ASUS M5A97 R2.0 2x4GB (40mm Noctua fan ziptied to VRM)
- FX-6300 running at 4.7GHz, 1.36V
- AMD R9 "Gamer Series" 2x4GB @ 2133MHz CL10, 1.65v
- Kingston SSDNow UV400 120GB
- XFX HD 7750 Ghost, 950C 1500M (modded VBIOS)
- Corsair CX600, ca. 2014
- Deepcool Gammaxx 400

 

[Retrorockit]

I did something today on my old Optiplex 380. It's running newer RAM on an older system. It's DDR3 but DDR2 works the same way.
My Opti 380 runs RAM at 1066 speed. No setting in the BIOS, no timings to tweak. Take it or leave it. So I bought some DDR3 1600 RAM for it.
2 slots only so 4GB was required, and the $9 each price for used Crucial Ballaistix didn't hurt either. But the CAS 9 @ 1600 got my attention. 11 is common at that speed, and CAS9 is common at 1333.
So what happens when you stick this in an underclocked situation? It runs at CAS 6-6-6-16@1066MHz. Much better than the 7-7-7-21 I had before. About 15% better. Nobody publishes this spec. but HWInfo64, or PC Wizard can uncover the timing tables for you. These would run CAS5 at 800 speed. A lot of DDR2 800 is CAS6. It has XMP profiles too but no way I know of to access them in a Dell Optiplex. This doesn't apply to overvolted RAM with low CAS#, watch out for that.

 

[Mr.Scott]

I did something today on my old Optiplex 380. It's running newer RAM on an older system. It's DDR3 but DDR2 works the same way.
My Opti 380 runs RAM at 1066 speed. No setting in the BIOS, no timings to tweak. Take it or leave it. So I bought some DDR3 1600 RAM for it.
2 slots only so 4GB was required, and the $9 each price for used Crucial Ballaistix didn't hurt either. But the CAS 9 @ 1600 got my attention. 11 is common at that speed, and CAS9 is common at 1333.
So what happens when you stick this in an underclocked situation? It runs at CAS 6-6-6-16@1066MHz. Much better than the 7-7-7-21 I had before. About 15% better. Nobody publishes this spec. but HWInfo64, or PC Wizard can uncover the timing tables for you. These would run CAS5 at 800 speed. A lot of DDR2 800 is CAS6. It has XMP profiles too but no way I know of to access them in a Dell Optiplex. This doesn't apply to overvolted RAM with low CAS#, watch out for that.

Ever heard of Memset?
Try it.

 

[Dinnercore]

I gave the P45 Platinum a small treatment. New pads and paste, the original red/purple paste on there was rock solid. While I was at it I also thought about a way to get voltage readings from boards without me having to hold my probes in place.
775 boards rarely offer any dedicated connectors for voltage measurement. Luckily I found cable connectors that my probes stick very well into.
I now have Vcore from the back of the socket, Vdimm from a choke on the RAM VRM, Vnb from a choke on the nb VRM and two separate grounds (one from the screw hole near the CPU power plug and one from the back of the socket).

Board is up and running after a few headaches. First I had a dead CPU (was a gift from someone who was not sure if it still works, but I only remembered that a few minutes later...). Next it seemed to struggle with my memory sticks. They were tested working, but the board got stuck cycling RAM detection.

Got it to boot finally, but it seems the MSI P45 behaves similiar to DFI boards when it comes to RAM. After trying a 2nd stick it cycled again but got stuck to the point where it would no longer post with any configuration. Single stick, any slot -> nope.
Turn off power at the PSU, remove battery, reset CMOS -> nope.
Turn off power at the PSU, wait 15 minutes (or until the DMM measures >0.0004V between ground and RAM) -> boots like nothing ever happened.

These LEDs are nearly bright enough to hurt. It's running now and I hope I can get a grip on the memory issues. It seems very aggressive with auto timings, it ran the single stick Cell Shock auto at 4-4-4-10-28 which is not even on the SPD for that stick.

 

[Retrorockit]

Ever heard of Memset?
Try it.

Memset looks interesting. But. It recognizes my DDR3 1600 as DDR2. It says MCHBAR is locked, which is really no surprise on a Dell. But the biggest problem is me. I've been flogging Dells for so long I have no idea how to use it! If I could find the XMP profiles I would have a starting point. I'll see if I can get it going. But Brave Browser doesn't have a translator yet. So the French website is not much help so far. But thanks I do like this type of program.

 

[Bones]

Memset looks interesting. But. It recognizes my DDR3 1600 as DDR2. It says MCHBAR is locked, which is really no surprise on a Dell. But the biggest problem is me. I've been flogging Dells for so long I have no idea how to use it! If I could find the XMP profiles I would have a starting point. I'll see if I can get it going. But Brave Browser doesn't have a translator yet. So the French website is not much help so far. But thanks I do like this type of program.

To me this sounds more like a Dell thing (BIOS) making it read that way.

Memset does work, allows timing adjustments on the fly in the OS so you don't have to reboot to tweak RAM timings in the BIOS. It's not perfect and in some cases doesn't work properly but in these cases it's normally as I'm thinking it is that causes the problem.

XMP is mainly for newer systems but there are some DDR3 setups/sticks that have it, there's just not too many that do and I'm not aware of many, if any 775 systems that have an XMP option in the BIOS.

Big thing is to figure out the starting point of your timings and work it from there, takes time but once you see gains it becomes easier to do as you go - Can't hurt to try it and see what happens.

Just don't count on radical gains like you see from actual OC'ing setups, your board/system is what it is so that's what you've got and have to work with I'm afraid. I will say so far you've done pretty good, so keep up the good work.

 

[Retrorockit]

The Crucial Ballisticx DDR3 in my Opti 380 has XMP. I tried to increase a setting from 6 to 7 and nothing moved. {t didn't even allow an entry. I've used Throttlestop to change Voltage and multiplier for years on Dells, so I'm not skeptical about it working. It just hasn't happened for me yet. I see it has winring0dll which is what Throttlestop uses to reach back to the BIOS form Windows. I didn't spend much time on it because I'm working an a different system right now. It's interesting but not at the top of my list. I don't expect the Dell to "HAVE" XMP, but if I knew the XMP settings I would have a starting point It might have to wait until I do an X58 project. maybe XMP needs an IMC to work?

 

[MachineLearning]

I want to preface this by saying I know I post a lot in this thread, lots of pictures which take up much room, so if it bothers people I'd be more than happy to make a different thread. Just let me know please, I want to post, not to take over.

Now, for more X58 adventures... This computer is a complimentary "sister" build to the other X58 one I posted about: same model case, similar motherboard, previously the same cooler (TR PA120), same CPU, same RAM - all the RAM even has sequential serial numbers. However, I ran into a heat wall when I wanted to bring a W3680 to the max, so that's where a new cooler - and newer CPU (2012 vs 2010 prod. date) come in.

Full specs:
- ASUS P6X58D Premium (favorite motherboard I have as of 10 minutes ago )
- Xeon X5675 (different sample than the previous build, but may be sister CPU)
- 3x4GB Kingston HyperX Fury DDR3 1866 @ 1.60v, 2090MHz, CL9-10-10-28-160-1T
- Thermalright Frost Commander 140 w/ MX-4 (holy moley this is supermassive... legitimately <=1mm of clearance to the glass!)
- ATI Radeon HD 5970 Reference @ stock
- Crucial MX500 1TB
- Super Flower Leadex SE 1000W Gold PSU
- Phanteks P400AD w/ 1x Arctic P12 PWM exhaust

Please ignore the cable spaghetti to the left, I swear I'll do better in the future

Also, HD 5970 needs some dusting, but the real star is this little Xeon...

Now, my last X5675 I didn't push very hard at all. It struck me as a good sample, and I'd rather keep as many CPUs in near-perfect condition as possible. But with these Xeons so cheap, and such a wonderful motherboard in the arsenal, of course I've gotta push something.

5.089GHz Suicide Run Validation
- this was the max I could attain on air, not even close to P95 stable. With a custom loop or a great AIO I think 4.8GHz+ is realistic, 1.5v+ just overloads seemingly any air cooler in P95.
4.614GHz Daily Validation
- 4.70GHz was just beyond reach. I needed more voltage, just air cooling can't handle it... maybe I'll try some MX-5 or something to see if that helps any?

The BCLK went higher than I assumed it would, I was booting close to 230BCLK (unstable due to the resulting core clock, didn't try to get it as high as I could). I think I conflated the approx. 205MHz BCLK "wall" on P55 with X58?
Anyway, I'm really happy about that 5+GHz run. That was exactly what I was hoping for, even if it's beyond unstable!

 

[Bones]

The Crucial Ballisticx DDR3 in my Opti 380 has XMP. I tried to increase a setting from 6 to 7 and nothing moved. {t didn't even allow an entry. I've used Throttlestop to change Voltage and multiplier for years on Dells, so I'm not skeptical about it working. It just hasn't happened for me yet. I see it has winring0dll which is what Throttlestop uses to reach back to the BIOS form Windows. I didn't spend much time on it because I'm working an a different system right now. It's interesting but not at the top of my list. I don't expect the Dell to "HAVE" XMP, but if I knew the XMP settings I would have a starting point It might have to wait until I do an X58 project. maybe XMP needs an IMC to work?

Use the SPD tabs in CPU-Z, it should show you all the settings for what speeds are programmed into the sticks.
You can use those values as a baseline and go from there.

 

[lexluthermiester]

maybe I'll try some MX-5 or something to see if that helps any?

Might help a little, but I think your limiting factor is the heat saturation capacity of the cooler itself, not the heat transfer aspect.

 

[Retrorockit]

Use the SPD tabs in CPU-Z, it should show you all the settings for what speeds are programmed into the sticks.
You can use those values as a baseline and go from there.

CPUZ SPD just shows the timings at the rated 1600 speed. Nothing for what's happening at 1066. PC Wizard is an old CPUID program that goes into the settings for all available speeds. HWInfo 64 does this also. For older hardware that's useful info. It even included OC timing for these modules. Up to about 1950 speed.

 

[Dinnercore]

CPUZ SPD just shows the timings at the rated 1600 speed.

In that case your sticks don't have SPD profiles for different speeds. CPU-Z does show them usually:

 

[Retrorockit]

That's not what I got from CPUZ SPD it had different timings all at 800 speed. including XMP. But just buying these and sticking them in the Dell I'm getting 6-6-6-16 at 1066 speed. CPUZ SPD doesn't have room for all the info. I'm talking about.
Here's a CPUZ Validation showing the timings I actually have. They don't show up in CPUZ SPD.

Intel Core 2 Quad Q9505S @ 2825.98 MHz - CPU-Z VALIDATOR

[74uad7] Validated Dump by 20DOLLARDELL (2022-02-03 15:44:41) - MB: Dell 0HN7XN - RAM: 8192 MB

valid.x86.fr

 

[stinger608]

I want to preface this by saying I know I post a lot in this thread, lots of pictures which take up much room, so if it bothers people I'd be more than happy to make a different thread. Just let me know please, I want to post, not to take over.

It's all good bro!!!!!!! Everyone loves the pictures I'm sure.

 

[Mr.Scott]

I want to preface this by saying I know I post a lot in this thread, lots of pictures which take up much room, so if it bothers people I'd be more than happy to make a different thread. Just let me know please, I want to post, not to take over.

Now, for more X58 adventures... This computer is a complimentary "sister" build to the other X58 one I posted about: same model case, similar motherboard, previously the same cooler (TR PA120), same CPU, same RAM - all the RAM even has sequential serial numbers. However, I ran into a heat wall when I wanted to bring a W3680 to the max, so that's where a new cooler - and newer CPU (2012 vs 2010 prod. date) come in.

Full specs:
- ASUS P6X58D Premium (favorite motherboard I have as of 10 minutes ago )
- Xeon X5675 (different sample than the previous build, but may be sister CPU)
- 3x4GB Kingston HyperX Fury DDR3 1866 @ 1.60v, 2090MHz, CL9-10-10-28-160-1T
- Thermalright Frost Commander 140 w/ MX-4 (holy moley this is supermassive... legitimately <=1mm of clearance to the glass!)
- ATI Radeon HD 5970 Reference @ stock
- Crucial MX500 1TB
- Super Flower Leadex SE 1000W Gold PSU
- Phanteks P400AD w/ 1x Arctic P12 PWM exhaust

View attachment 235082
View attachment 235083
Please ignore the cable spaghetti to the left, I swear I'll do better in the future
View attachment 235084
View attachment 235085
Also, HD 5970 needs some dusting, but the real star is this little Xeon...

View attachment 235087View attachment 235088

Now, my last X5675 I didn't push very hard at all. It struck me as a good sample, and I'd rather keep as many CPUs in near-perfect condition as possible. But with these Xeons so cheap, and such a wonderful motherboard in the arsenal, of course I've gotta push something.

5.089GHz Suicide Run Validation
- this was the max I could attain on air, not even close to P95 stable. With a custom loop or a great AIO I think 4.8GHz+ is realistic, 1.5v+ just overloads seemingly any air cooler in P95.
4.614GHz Daily Validation
- 4.70GHz was just beyond reach. I needed more voltage, just air cooling can't handle it... maybe I'll try some MX-5 or something to see if that helps any?

The BCLK went higher than I assumed it would, I was booting close to 230BCLK (unstable due to the resulting core clock, didn't try to get it as high as I could). I think I conflated the approx. 205MHz BCLK "wall" on P55 with X58?
Anyway, I'm really happy about that 5+GHz run. That was exactly what I was hoping for, even if it's beyond unstable!

Love your pics and info. Keep it coming. That's what this threads all about.
BTW, have you increased your PCIe bus for high bclk stability? Bump that to 110MHz and test. You can go to about 120MHz if you have to.

 

[Susquehannock]

I want to preface this by saying I know I post a lot in this thread, lots of pictures which take up much room, so if it bothers people I'd be more than happy to make a different thread. Just let me know please, I want to post, not to take over.
<snip>

x2 on what Mr Scott just said. Know I don't post in this thread much. but I like the pictures.

 

[Retrorockit]

If it wasn't for the photos i wouldn't know what most of this tsuff is.

 

[Bones]

That's not what I got from CPUZ SPD it had different timings all at 800 speed. including XMP. But just buying these and sticking them in the Dell I'm getting 6-6-6-16 at 1066 speed. CPUZ SPD doesn't have room for all the info. I'm talking about.
Here's a CPUZ Validation showing the timings I actually have. They don't show up in CPUZ SPD.

Intel Core 2 Quad Q9505S @ 2825.98 MHz - CPU-Z VALIDATOR

[74uad7] Validated Dump by 20DOLLARDELL (2022-02-03 15:44:41) - MB: Dell 0HN7XN - RAM: 8192 MB

valid.x86.fr

We get that, note I said to use these values a a baseline to tweak from.
The timings shown using Dinnercore's example:

In that case your sticks don't have SPD profiles for different speeds. CPU-Z does show them usually:

Says for XMP-1600 timings of 9-9-9-24 are for running the sticks at 1600 BUT that doesn't mean you'd always use those timings/specs to go higher - You can certainly try it but eventually you run into the limit of what the sticks are capable of.
Only real way to continue is of course to loosen timings, maybe bump RAM voltage a bit and go from there.

Also bear in mind the values such as 593, 667 and so on as shown in the columns you see are based on DDR (Double Data Rate) - That means the actual speeds are doubled for what's shown. That's why beneath where it's saying "XMP-1600" you see 800MHz as the value given in that slot (800 X2 = 1600) so the value(s) you're looking for are there, just need to interprit them as outlined and go with it.

Hope this helps.

 

[Keullo-e]

I want to preface this by saying I know I post a lot in this thread, lots of pictures which take up much room, so if it bothers people I'd be more than happy to make a different thread. Just let me know please, I want to post, not to take over.

Now, for more X58 adventures... This computer is a complimentary "sister" build to the other X58 one I posted about: same model case, similar motherboard, previously the same cooler (TR PA120), same CPU, same RAM - all the RAM even has sequential serial numbers. However, I ran into a heat wall when I wanted to bring a W3680 to the max, so that's where a new cooler - and newer CPU (2012 vs 2010 prod. date) come in.

Full specs:
- ASUS P6X58D Premium (favorite motherboard I have as of 10 minutes ago )
- Xeon X5675 (different sample than the previous build, but may be sister CPU)
- 3x4GB Kingston HyperX Fury DDR3 1866 @ 1.60v, 2090MHz, CL9-10-10-28-160-1T
- Thermalright Frost Commander 140 w/ MX-4 (holy moley this is supermassive... legitimately <=1mm of clearance to the glass!)
- ATI Radeon HD 5970 Reference @ stock
- Crucial MX500 1TB
- Super Flower Leadex SE 1000W Gold PSU
- Phanteks P400AD w/ 1x Arctic P12 PWM exhaust

View attachment 235082
View attachment 235083
Please ignore the cable spaghetti to the left, I swear I'll do better in the future
View attachment 235084
View attachment 235085
Also, HD 5970 needs some dusting, but the real star is this little Xeon...

View attachment 235087View attachment 235088

Now, my last X5675 I didn't push very hard at all. It struck me as a good sample, and I'd rather keep as many CPUs in near-perfect condition as possible. But with these Xeons so cheap, and such a wonderful motherboard in the arsenal, of course I've gotta push something.

5.089GHz Suicide Run Validation
- this was the max I could attain on air, not even close to P95 stable. With a custom loop or a great AIO I think 4.8GHz+ is realistic, 1.5v+ just overloads seemingly any air cooler in P95.
4.614GHz Daily Validation
- 4.70GHz was just beyond reach. I needed more voltage, just air cooling can't handle it... maybe I'll try some MX-5 or something to see if that helps any?

The BCLK went higher than I assumed it would, I was booting close to 230BCLK (unstable due to the resulting core clock, didn't try to get it as high as I could). I think I conflated the approx. 205MHz BCLK "wall" on P55 with X58?
Anyway, I'm really happy about that 5+GHz run. That was exactly what I was hoping for, even if it's beyond unstable!

That is a great board. I have the cheaper P6X58D-E and it's simply an awesome 1366 board.

 

[Retrorockit]

You don't just "bump" RAM Voltage on a Dell. The BIOS has absolutely no settings except boot sequence.
This was very good simple mod for that type of computer. It is what it is. It doesn't matter what software you look at it with. It's not going to change.
Memset may or may not work out. But right now I've Got an XPS430 running 4.15GHz on a QX9650, and GTX1660Ti. I'm looking to add 2x 120GB SSDs in RAID0.
I want to get 4.33Ghz if I can. The next step for the Opti 380 is to see if i can get a BIOS that will allow a QX9650 to run in it. X5470 Xeon swap-Been there, done that. Tweaking the RAM timings further is not what I'm really interested in.

 

[MachineLearning]

It's all good bro!!!!!!! Everyone loves the pictures I'm sure.

Love your pics and info. Keep it coming. That's what this threads all about.
BTW, have you increased your PCIe bus for high bclk stability? Bump that to 110MHz and test. You can go to about 120MHz if you have to.

x2 on what Mr Scott just said. Know I don't post in this thread much. but I like the pictures.

Thank you all I'm thinking of doing roughly 1 post / feature per day here, to pace myself and help keep this thread lively.

On the PCIe bus: I've had either crap luck with it helping, or something in my specific configuration limits it from letting me raise BCLK higher. I just tested with low RAM / core speed, and wasn't able to beat 221 BCLK with either 112MHz PCIe (max this system seems to like) or flat 100MHz PCIe. Perhaps I'm missing something, or maybe I need to disable some devices like the Marvell SATA III controller (connected via PCIe x1)?

And, here's a XFX HD 4890 for y'all...

We'll see how far I can bring her... Hoping for 1GHz core w/ increased voltage, no idea what the memory should attain. However I definitely am very happy to finally get some Infineon / Qimonda VRAM!

As you can somewhat see in the last photo, the PCB is bent a bit upwards towards the end of it. I received it that way (sans electrical tape), but it works so I don't mind. I taped it up to stop it from eventually splintering or suffering some more damage. It looks like someone bumped it forcefully on a flat surface so if that ever happens on my watch, there's now a buffer...

Also something to note, this is not the XFX Black edition (1GHz core/mem @ stock), but it does have an 8+6pin PCIe input as opposed to the reference 6+6pin PCIe. It still looks like an identical PCB otherwise (barring color / minor component choices, like that inductor in the topmost right of the front XFX PCB).

 

[Retrorockit]

I fired up my Dell XPS420. I've got the QX9650 running 4.32GHz. It's holding 70* with just a dry thermal pad for TIM. The 420 wasn't one of Dells big gaming rigs, it was a mid range multimedia system.
But it's locked BIOS, BTX, and DDR2 12GB. Because it's BTX no aftermarket coolers. The air cooling is cobbled together out of Dell workstation, and XPS parts. IDK if I can squeeze more out of it. But I've got a few tricks up my sleeve. I'm faking SATA3 with a 2x120GB SSD RIAD0 setup courtesy of the Dell BIOS.
My CPUZ was 270, and 1049. Now it's 390 and 1495.

 

[Keullo-e]

Thank you all I'm thinking of doing roughly 1 post / feature per day here, to pace myself and help keep this thread lively.

On the PCIe bus: I've had either crap luck with it helping, or something in my specific configuration limits it from letting me raise BCLK higher. I just tested with low RAM / core speed, and wasn't able to beat 221 BCLK with either 112MHz PCIe (max this system seems to like) or flat 100MHz PCIe. Perhaps I'm missing something, or maybe I need to disable some devices like the Marvell SATA III controller (connected via PCIe x1)?

And, here's a XFX HD 4890 for y'all...

View attachment 235134
View attachment 235124
View attachment 235125
View attachment 235129
View attachment 235126
View attachment 235127

We'll see how far I can bring her... Hoping for 1GHz core w/ increased voltage, no idea what the memory should attain. However I definitely am very happy to finally get some Infineon / Qimonda VRAM!

As you can somewhat see in the last photo, the PCB is bent a bit upwards towards the end of it. I received it that way (sans electrical tape), but it works so I don't mind. I taped it up to stop it from eventually splintering or suffering some more damage. It looks like someone bumped it forcefully on a flat surface so if that ever happens on my watch, there's now a buffer...

Also something to note, this is not the XFX Black edition (1GHz core/mem @ stock), but it does have an 8+6pin PCIe input as opposed to the reference 6+6pin PCIe. It still looks like an identical PCB otherwise (barring color / minor component choices, like that inductor in the topmost right of the front XFX PCB).

How about two of them?

 

[MachineLearning]

How about two of them?

That's awesome! I thought they were identical at first glance, what's up with the different fan and sticker? Is that your own replacement or another revision? I couldn't find anything on google although maybe didn't look long enough

 

[Keullo-e]

That's awesome! I thought they were identical at first glance, what's up with the different fan and sticker? Is that your own replacement or another revision? I couldn't find anything on google although maybe didn't look long enough

Another revision I guess, it also has a slightly different PCB. It had that cooler when I bought it (and the other one) used.