TPU's Nostalgic Hardware Club

[RyzenMaster.sys]

Yeah those are kind of monitor I was looking for initially but then this came along and I'm really sucker for EIZO monitor as I have Eizo Flexscan E55D before, even though its only 15 inch but it uses Sony aperture grill, image quality is simply amazing.

I might have to be on the lookout for one sometime. Most brands are hit or miss unfortunately, so its nice to see a brand that makes mostly great monitors. On that note I do truly think that Dells monitors are the best product they manufacture in modern era.

 

[Greenslade]

I was just looking at this video from one of my fav Retro YouTuber,s

and decided to look on Facebook on market place and saw this for £73 ?

CyberpowerPC Wyvern Gaming PC – Intel Core i5-10400F, Nvidia GTX 1650 4GB, 16GB £73 ?New conditionThats must be a fake add. I would give him that just for the GPU.The Retro you tuber had another video on the same modal as the one in the video ,that he also got for 10 pounds, of Facebook Market place but was working.

 

[lexluthermiester]

Thats must be a fake add.

Very likely a scam. Don't waste your time..

 

[Greenslade]

Very likely a scam. Don't waste your time..

It has to be, and no i want You don,t get anything for nothing in this life.I was hoping i could get a bargin like the guy in the video.The one he has in the video are 25 pounds cheapest i have seen them on eBay, and they only have E 4600 in them and no GPU.

Very likely a scam. Don't waste your time..

This is the one i could not find

. Trying it with a new Gpu.

 

[RyzenMaster.sys]

HIS Radeon X1650 Pro 512mb

I wanted to build a socket 939 system around a year ago. I picked up an Asus A8N SLI Premium with an Athlon 64 X2 4400+ atop it. However the board doesn't function properly. Eventually I'll find a nice board and finish the system.

 

[MachineLearning]

View attachment Screenshot (17).png

CPU-Z Validation

Got around to overclocking my unlockable Phenom II 960T on my previously featured ASUS M4A89GTD Pro/USB3. Voltage regulation on this board isn't the worst I've seen, but I wouldn't call it great. However I suppose this is one of the higher-power draw scenarios that this VRM may ever face (highest would prob be an FX-8100) so that's worth considering. vCore fluctuated by about 40mv during testing, at idle it was typically closer to 1.535-1.55v and under load 1.55-1.575 (with approx. 1.575v set in BIOS, & LLC enabled). CPU/NB was set to 1.45v, this sample hates 2800MHz uncore so I kept it around 2700MHz. The best I achieved was 4.177GHz on all 6 cores, not too shabby. There are two problems that I have with the overclocking on this board:

1. BCLK is absolutely never what you set, under any circumstance. Yes, spread spectrum is turned off for all applicable components... and it's far more severe than that anyway, deviating by anywhere from +4MHz to +0.7MHz BCLK (power cycle causes the deviation to shift).
2. vCore and CPU/NB voltages are set via offset, no option to change that. Annoying, but the BIOS at least will tell you what voltage you're currently at so you can go from there.
~
I also cleaned up, repasted and overclocked a bit using this ASUS M4N98TD Evo, but was confronted with a bigger roadblock than on the board before, that being very poor voltage regulation. It's a keeper still, this has the 980a chipset which was nVidia's last for AMD, and as far as I can tell, overall. It's basically the 780a rebranded, main difference being that most boards with the 980a use DDR3.

This is as the board came in, with some dust across the board that may not be easily visible on camera and what I'm pretty sure are Noctua AM3 mounting brackets. Also a small detail, it seems the previous owner decided the CPU fan header needed to be reversible (??), so the plastic guide is missing there.

There is a highly technical term I've heard a couple of times to describe the NB temperature in this picture: "not good."

Well, here's why. Whatever thermal interface once existed is now a dry yellow splotch, and will not come off in this century. I put MX-2 on top of it in the hopes that it'd suffice, and do that it did!

Ahhh... much better.

Marketing claims that this is a 8-phase, I personally think this is a 4+1 phase VRM with 2 high, 2 low side MOSFETs, 2 inductors/phase? I don't fully understand VRMs, this is one of my weakest areas, so if anyone knows what they're talking about I'd love to hear your opinion on this.

As it came out. I stole some AM3 brackets from my nuked MSI VRM fail edition 970 board I assume that stock, it would have had the classic round mount. I gave the whole surface a light brushing, and blasted it with an air duster front and back, plus the heatsink assembly separately. A couple fins on the VRM were very slightly bent, so I tried to straighten those out too.

This board refuses to post with any 8GB DIMM that I have, so I've been running it mainly with my Kingston HyperX Genesis 2x4GB kit.

View attachment Screenshot (21).png

CPU-Z Validation

For this motherboard, I decided to try and unlock my Phenom II 555 BE. Indeed, this board was able to get all 4 physical cores up and running. One small annoyance is that while this board does have NVCC and therefore core unlock, you can't choose which cores you want active, it's all of them or stock only. Core #3 is the worst on this CPU, always giving up earliest, so I would have liked to disable it to try for a 3-core higher OC. The M4A89GTD Pro/USB3 allows per-core enabling/disabling in BIOS.

Voltage regulation, as I mentioned before, is bad. I started out with that same PhII 960T BE as the beginning of this post, and voltage was immediately -25mv from what I set, and then -60mv from there. Rough. Then of course, the Phenom II X4 B55, which I hoped would be a bit better... nope, in fact it's virtually the same. The above image is with 1.65v vCore set, 1.5v CPU/NB, 1.4v LDT/HT Voltage, stock 980a chipset and NF200 voltages (1.1v and 1.2v respectively). This sample seems to be kind of crappy, even though it's a C3 revision all 4 cores top out at 3.8GHz (4.0GHz was bootable, but quickly problematic). I bet that with disabling core #3 on a board with better voltage regulation, 4.0GHz stable might be within reach for this CPU.

Despite my complaints, again, this board is a keeper, if only for that low-effort 980a...

 

[Dinnercore]

@MachineLearning I have faced this dried out yellow stuff a few times already on chipsets. It's really horrible to get off, Isopropyl does nothing against it. I resort to mechanical removal with a very fine blade.

On another note, I've been binning and benching the first 7 of my 21 E8500s. I'm fascinated by the extreme differences in the properties of each individual CPU. Not just the frequency capabilities, but power consumption and heat at a fixed given voltage.
The max. frequency @1.35V (measured real voltage, not bios or CPU-Z value) ranges from 4.08GHz (C0 stepping sample) to 4.5GHz. As expected. But very interesting is that one of the very low VID CPUs shows an extreme case of leakage? Or whatever the technical term may be. At just 1.25V (the board with enabled LLC actually overshoots the voltage by 5-8mV in IDLE) the system draws 120W idle. Where other CPUs idled around 90W @ 1.35V. As expected it also heats up a lot more. Load power draw is already at 164W just at 1.25V.
Just by increasing the frequency 200MHz on that sample, the temps went up 4-5°C on the same voltage. I'm kinda curious to see how this chip will perform under cold. 300W C2D? Or does the cold reduce the leakage to more reasonable levels? I'll try to find out.

 

[MachineLearning]

@MachineLearning I have faced this dried out yellow stuff a few times already on chipsets. It's really horrible to get off, Isopropyl does nothing against it. I resort to mechanical removal with a very fine blade.

On another note, I've been binning and benching the first 7 of my 21 E8500s. I'm fascinated by the extreme differences in the properties of each individual CPU. Not just the frequency capabilities, but power consumption and heat at a fixed given voltage.
The max. frequency @1.35V (measured real voltage, not bios or CPU-Z value) ranges from 4.08GHz (C0 stepping sample) to 4.5GHz. As expected. But very interesting is that one of the very low VID CPUs shows an extreme case of leakage? Or whatever the technical term may be. At just 1.25V (the board with enabled LLC actually overshoots the voltage by 5-8mV in IDLE) the system draws 120W idle. Where other CPUs idled around 90W @ 1.35V. As expected it also heats up a lot more. Load power draw is already at 164W just at 1.25V.
Just by increasing the frequency 200MHz on that sample, the temps went up 4-5°C on the same voltage. I'm kinda curious to see how this chip will perform under cold. 300W C2D? Or does the cold reduce the leakage to more reasonable levels? I'll try to find out.

The only thing worse I've seen is this deep red "paste" some boards used, it basically welds the chipset and heatsink together and it's terrifying to take off. This yellow crap is definitely second worst, though. .

I'd love to see your E8500 findings and compare them to my 6, I have a much lower sample size and we won't have the same setup but surely something can be extrapolated. I also saw massive power draw differences at identical settings, the difference between samples is insane sometimes.

 

[Dinnercore]

The only thing worse I've seen is this deep red "paste" some boards used, it basically welds the chipset and heatsink together and it's terrifying to take off. This yellow crap is definitely second worst, though. .

Just had that on the SB of my Rampage III Extreme. In these cases its probably enough to just throw some paste on top, but I always go through the pain and clean it off. I simply have to do it

 

[lexluthermiester]

Marketing claims that this is a 8-phase, I personally think this is a 4+1 phase VRM with 2 high, 2 low side MOSFETs, 2 inductors/phase?

Nope, it's 8 phase. 2 mosfets per phase with a choke each. The extra choke is likely a line conditioner for the input voltage.

 

[Ruru]

Some boards use doublers and market their VRM as the doubled amounts. Dunno more, I don't know much about VRMs but that's what I remember reading.

2nd PC with CPU & GPU watercooled

I'll need to check how the VRM/VRAM manages or do I need to search for the PCI-slot fan bracket for blowing some air to that heatsink..

edit: gets hot but not unbearable hot during 3dmark, I guess it's fine for gaming as games doesn't stress the card as much.

3DM FS: https://www.3dmark.com/fs/27291076
3DM TS: https://www.3dmark.com/spy/26580624

 

[lexluthermiester]

Some boards use doublers and market their VRM as the doubled amounts. Dunno more, I don't know much about VRMs but that's what I remember reading.

While that did happen, the circuit pathways tell the story on that board. Those traces are running parallel instead of in series, with directly indicates 8 phases instead of 4 phases doubled up.

 

[Ruru]

While that did happen, the circuit pathways tell the story on that board. Those traces are running parallel instead of in series, with directly indicates 8 phases instead of 4 phases doubled up.

I mean that some people (like me) with less knowledge could fall to that marketing bullshit. Ah, good old times.

Luckily I have good boards (P5Q Pro and this P6X58D-E) for some serious 775 and 1366 overclocking.

 

[lexluthermiester]

I mean that some people (like me) with less knowledge could fall to that marketing bullshit. Ah, good old times.

To be fair, whether it's true 8 phase or 4 phase with dual line regulation, the overall effect is similar, quality power delivery. So in the end it would be ok either way as long as quality parts were used and proper engineering done.

 

[Ruru]

To be fair, whether it's true 8 phase or 4 phase with dual line regulation, with overall effect is similar, quality power delivery. So in the end it would be ok either way as long as quality parts were used and proper engineering done.

Yeah and the components in the VRM part matter too. Just like with capacitors.

 

[RyzenMaster.sys]

Asus baffles me.

They make most of their mid range or high end normal board for 1366 have 16 phase vrms (really 8 with doublers iirc). Only the cheapest models they make for 1366 have 8 phases like the Asus P6T.

So they make the Rampage III Formula with all of these extreme overclocking features and ln2 mode etc... and what do they use for the vrm? 8 phases. With similar rated power stages to the mid range 16 phase boards not intended for such overclocking.

Asus used 16 phases (usually 8 doubled) basically at any chance they could get back then. My Asus P8Z68-V Deluxe Gen3 has a 16 phase vrm too. They even put it on LGA 775 with boards like the P5Q Deluxe.

 

[Ruru]

Edited the Fire Strike and Time Spy results of that setup few posts above.

 

[lexluthermiester]

Got a Q9550 and Radeon R7-250 ordered for the Lenovo I discussed previously. I will post more pictures and screenshots of performance results. Not expecting blazing fast performance because, let's be honest, it's a Q9550 and a Radeon R7-250 2GB. Yet it is a small form factor PC, so if it does well enough, I'll be happy. If not, I'm going to hunt down a GTX750...

EDIT: Low-profile GTX-750..

 

[Dinnercore]

I may have gotten really lucky. This one low VID E8500 out of the recently tested bunch could be worthy of the 'golden' sample title.

Batch is Q836A222.

Passed R15 with 1.25V @ 4.50GHz. Probably not prime stable as R15 is not the most demanding bench. I reserve further testing for subzero cooling.

Now the batch is where things get interesting. I have another one (not tested yet) on my tray of 21 that matches the exact batch number:

Not sure how much truth there is to CPUs from the same batch performing similar. I would imagine defects occur at random across all wafers in a kind of equal fashion, so there should be the same differences between CPUs from the same batch as with CPUs from other batches. But who knows, maybe on that fateful day/week the environmental conditions inside the cleanroom were better compared to other days.

Also on top of these two with identical batch, I have three more from Q836AXXX. All three are Q836A079. None of them tested yet.

The latest batch I found on my E8500s is Q114E134, which should be from 2011 if I'm not mistaken.

While I was at it, here is a picture of the first 'Intel Confidential' in my collection. Pentium 4 550 qualification sample.

From 9 binned E8500s so far, my worst one is of course the C0 stepping. 4.08GHz @ 1.35V. Worst E0 out of the bunch does 4.20GHz @ 1.35V. The rest of them fall within 4.30GHz - 4.40GHz @ 1.35V.

 

[seth1911]

@Dinnercore

Make R20 plese, im sure my A10 7870k will be slower


Edit: No Comment (FF in background cause download) (Celeron G6900@ 5,35 GHz 1449)
So my new 10400F will have on 1 core an apu from 2016

 

[Dinnercore]

Make R20 plese, im sure my A10 7870k will be slower

Can we meet at R23? Don't have R20 on my Win 10 ssd yet. Maybe I'll install it later between tests, but no promise.

This is one of the worst E0 samples. Needs a lot of voltage to get going, but it has very low power consumption and heat at the same time. Still water cooled and ~62°C under load.

 

[seth1911]

in 1 Thread: A A10 7870k fcked from a Core 2 Duo

 

[Trekkie4]

Heh, if I only had this guy's budget to spend on a sleeper build... Still, pretty awesome idea if I may say so!

 

[Ahhzz]

Can't seem to figure out why this disc won't fit in the DVD drive.....



And for my next trick....



btw, I will be posting up a small SCSI collection and the WinTV cards in the {FF} section a little later if any of you nostalgic types are looking for anything stupid old heheh

 

[Trekkie4]

Can't seem to figure out why this disc won't fit in the DVD drive.....

View attachment 238098

And for my next trick....
View attachment 238099View attachment 238100View attachment 238101


btw, I will be posting up a small SCSI collection and the WinTV cards in the {FF} section a little later if any of you nostalgic types are looking for anything stupid old heheh

Huh... This WinTV card of yours looks like Hauppauge but I can't seem to find their logo (or signature) anywhere.