Casanova

Well, it's been a while since last time i updated this thread with good old articles and sources about prehistoric ln2 overclock. Also, with the migration to this new forum, this thread lost almost all the pictures i attached (the links are still ok)... I wish i had more time to spare doing this researchs, so if anyone else is interested in continuing, here are the tips to find things on the web: 1 - web.archive.org - try all the search filters this website provides; 2 - google in japanese, using the date tool, to try find something from the 90s; Good luck!

Thanks, just another question: does he used a custom cpuz, or that's what cpuz reads when using x299/i9?

Amazing, congrats 2.256v....??

The guy is from extreme league, 1.71v is quite high for ambient temp, unless he benched in a room with near 0°C temp.

Volta is really crushing the pascal results.

This card is a beast! I'm really impressed, huge gains against Pascal tecnology. Thanks for sharing

LoL I like to do this type of adaptors too. Btw, great run! Congrats

Same here

Thank you for your dedication, downloading it now

Thanks mate Thanks. Ben told me about this, next time i'll try the non masochism method... As you wish, plus burned mosfet Thanks! Samsung mem chips.

Yes, but i'll wait to see which card i going to be pick before buying another gt 1030. Thanks, the photos are not very good because i'm also a noob photographer

First I want to thanks two great overclockers from our community that helped me: Ben @newlife and Alex @DarkVenom, thank you guys, i hope someday I’ll be in your level of mods and overclock skills. This thread is about my researches and experiences with Galax GT 1030 Vmod's. I started to study the card's PCB. Galax made an amazing card, and despite of this one being an entry level it already has vcheck points identified in the pcb. I also looked at the main VRM components of it, and, after a while i found that the IC uP9305W is responsible for the 2GB GDDR5 voltage control. https://www.upi-semi.com/en-article-upi-473-1459 At first, i thought that the IC uP1666Q was making both vgpu and vmem control, but after taking some multimeter measurements i realized that this IC was not being used for vmem. Taking a look at the uP9305W datasheet and testing the card with a multimeter it became clear that the FB (pin 6) was the one needed for the memory vmod. For this one I used a 50k trimpot, that gives the overclocker a more accurate and secure regulation of the vmem. Tested it, worked like a charm 😊 Now about the IC uP1666Q if you google for it you'll find that this IC is widely used in many other pascal cards, from 1060 to 1080ti's. https://www.upi-semi.com/en-article-upi-362-1566 As you can see from the datasheet, this IC is more complex than the uP9305W. After a couple hours testing it with a multimeter, i noticed that the pin 11 was the one needed for gpu vmod. The problem here is that it is so tiny that it was hard to wire it, but after sometime i managed to weld it. Unfortunately, i made a mistake of soldering one of the 20k trimpot legs to the wire just above the card, and a little drop of weld fall in a small smd resistor around the uP1666Q area, and i did not saw it .... the result? Fireworks in one of the mosfets; (.) The short was so strong that ripped one of the tracks! I'll buy another one next month and try to not kill it again So that's it, i wanted to share my experiences with this great card, any questions and critics to this vmod noob that write to you are welcome 😊

Guys, i could not find the datasheet of this specific IC, uP9305W PGA6BM. If anyone here has this datasheet, could you share it with me, please? Thanks in advance

Thanks mate, i really appreciate that :celebration: This thread turned out to be a great extreme overclock compendium, something that I think is very important for the overclock history. I'm almost sure that the first extreme LN2 overclock sessions were made in the early 90's. The oldest record I found was a video dating back to 1984 but it is a pioneering experiment done in a scientific environment, and i think it´s very unlikely that during the 80's someone performed this procedure at home as an enthusiast, but in the early 90's it is quite likely, and i'm almost sure it happened in Japan.

Ok, now this one is gold OC Archeology After some research, i found this japanese overclock website still online: FF page PC index FF page Celeronã‚’æ¶²ä½“çª’ç´ ã§å†·ã‚„ã™ It has all the overclock experiments made by a unknown overclocker, since 08/06/1996. This one in particular was made with LN2 on 09/23/1998, using 10 liters of this precious liquid to cool down a Celeron 333. The overclocker did a complete description about the overclocking session, including many photos and results of the experiment, which was very rare for the year 1998 and mostly without any reference to be followed. We can even say that this is certainly one of the forefront extreme overclock enthusiast with LN2 documented. To get an idea, the overclocker built a cpu pot with welded copper sheets!

Very well written and explained article, EVGA EPOWER V is a piece of art, congrats TiN! But, that price... ;(

Indeed, this slow loading is more noticiable on hwbot.org.

I digged this one from a Slovak website, dated from march 2002. Extreme ln2 overclock with a "brand new released" Northwood P4 Tunned hardware: Motherboard ABIT TH7, RIMM PC800 2x128MB 2.75V, cooling LN2 -184 C, CPU voltage 2.05V, Graphics card FVD4500PCI, SCSI Dc390u3d. link: Pretaktovaný Intel Pentium 4 2,4 Ghz (Northwood) | pc.sk

Guys, when i try to upload my GPUPI 1B for Road to Pro Division II round 3 it says in limitations that i have to make a submission for 3dmark03... wtf?? Follow print proof. @websmile @Massman any ideas? Or am i doing something wrong?

I had the same problem, just sent an email to @websmile with the print and archive.

Please disregard, it was me doing it wrong upload ok.

Good night! I successfully ran GPUPI 1B and 100M, printed according to the rules, archive submission is OK, but when i try to submit it gives the error "Error client version 2.3". Follow the google drive archives as prove. @Massman https://drive.google.com/open?id=0B7HP4KK3cTXGQ0EyelN0clU3Rk0 Can you guys correct it please? Thanks in advance.

I found this old 3d benchmark dated from 1997, released by remedy. Bunny's Workshop used this bench to competitions in the late 90's. The benchmark gives a final score. This one is not on Hwbot 3d bechmarks list, but i think it's worth be in there, and it can be used for Old School is the Best School competitions as well. Download link: https://files.scene.org/view/demos/groups/remedy/fr101.exe Bunny's Workshop ranks with Matrox Millennium G200 and other vintage graphics cards. Bunny's Workshop

These ones are also from the same website rankings for superpi1m pifast

@Massman I do not know if you've already cataloged these: superpi