Youll have to fight the P3 Tually over it .. it runs circles around the P4 ...even with a GeForce 3 ti500 the P3 handily beats it, Intel really did shit the bed with Netburst and its super long pipeline. The only case where the P4 manages to show its strength is when you run software than can saturate the DDR bandwidth on the P3, the P4 simply has more but for gaming . .its no contest the P3 beats it.

Now I wonder if the 1.7 P4 will fare better, they are pretty cheap on evilbay so I may grab one to see if the speed bump helps 😁 (The Tually can OC to 1.6 so itll still be fairly even, I think with a bit more tweaking and cooling I might get it a bit higher)

TrashPanda wrote on 2022-06-17, 19:37:

Intel really did shit the bed with Netburst and its super long pipeline.

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations killed that idea.

The only case where the P4 manages to show its strength is when you run software than can saturate the DDR bandwidth on the P3, the P4 simply has more but for gaming . .its no contest the P3 beats it.

Another case where the P4 really shows its strength is when SSE2 instructions are used and that was mainly with video encoders. On the other hand, NetBurst was inefficient in terms of performance per watt (and got much worse with Prescott) and software required specific optimizations to avoid the pitfalls.

The troubles with power consumption and heat dissipation led to the creation of the Pentium M (itself a derivative of the P6 microarchitecture) which paved the way for the Core microarchitecture years later.

Gmlb256 wrote on 2022-06-17, 20:12:

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations kille […]
Show full quote

TrashPanda wrote on 2022-06-17, 19:37:

Intel really did shit the bed with Netburst and its super long pipeline.

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations killed that idea.

The only case where the P4 manages to show its strength is when you run software than can saturate the DDR bandwidth on the P3, the P4 simply has more but for gaming . .its no contest the P3 beats it.

Another case where the P4 really shows its strength is when SSE2 instructions are used and that was mainly with video encoders. On the other hand, NetBurst was inefficient in terms of performance per watt (and got much worse with Prescott) and software required specific optimizations to avoid the pitfalls.

The troubles with power consumption and heat dissipation led to the creation of the Pentium M (itself a derivative of the P6 microarchitecture) which paved the way for the Core microarchitecture years later.

Core also includes a lot of the great things that Netburst had, not all of it was terrible so they really merged the P3/M core with the best of Netburst.

TrashPanda wrote on 2022-06-17, 20:14:

Gmlb256 wrote on 2022-06-17, 20:12:

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations kille […]
Show full quote

TrashPanda wrote on 2022-06-17, 19:37:

Intel really did shit the bed with Netburst and its super long pipeline.

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations killed that idea.

The only case where the P4 manages to show its strength is when you run software than can saturate the DDR bandwidth on the P3, the P4 simply has more but for gaming . .its no contest the P3 beats it.

Another case where the P4 really shows its strength is when SSE2 instructions are used and that was mainly with video encoders. On the other hand, NetBurst was inefficient in terms of performance per watt (and got much worse with Prescott) and software required specific optimizations to avoid the pitfalls.

The troubles with power consumption and heat dissipation led to the creation of the Pentium M (itself a derivative of the P6 microarchitecture) which paved the way for the Core microarchitecture years later.

Core also includes a lot of the great things that Netburst had, not all of it was terrible so they really merged the P3/M core with the best of Netburst.

True, SSE2 instruction usage became more common later on (especially with 64-bit OSes) and Nehalem microarchitecture did bring back Hyper-Threading.

Gmlb256 wrote on 2022-06-17, 20:33:

TrashPanda wrote on 2022-06-17, 20:14:

Gmlb256 wrote on 2022-06-17, 20:12:

Yep, Intel though that they could make a CPU that can scale to 10 GHz with those deeper pipelines but physical limitations killed that idea.

Another case where the P4 really shows its strength is when SSE2 instructions are used and that was mainly with video encoders. On the other hand, NetBurst was inefficient in terms of performance per watt (and got much worse with Prescott) and software required specific optimizations to avoid the pitfalls.

The troubles with power consumption and heat dissipation led to the creation of the Pentium M (itself a derivative of the P6 microarchitecture) which paved the way for the Core microarchitecture years later.

Core also includes a lot of the great things that Netburst had, not all of it was terrible so they really merged the P3/M core with the best of Netburst.

True, SSE2 instruction usage became more common later on (especially with 64-bit OSes) and Nehalem microarchitecture did bring back Hyper-Threading.

There was also a ton of optimizations built into the netburst Uarc that they moved over, its highly amusing that they had to go back to move forwards, in hindsight they should have just stuck with the P3 architecture to begin with.

Radical Vision wrote on 2022-06-17, 19:36:

So as i found that the temperatures on the Voodoo V 5500 chips are dumb high (specially the primary chip closest to the VGA port is really burning fingers....) and i never like to use hardware that cook itself slowly over the years (sure these cards working fine like that, but still they may die at some point from overheating..), so i plan to freeze the card in freezer and then to pop off the garbage joke of a heatsinks and replace them with at least 5 times bigger heatsinks and fans, or maybe even Zalman coolers.

I used my freezer and a nylon spudger to remove the absolutely cemented-on heatsink of a Radeon 8500 (don't ever use a screwdriver). It took two or three freezes until it finally broke off with a satisfying crack. Graphics cards have a lot of copper, and although it takes a long time to cool, they will heat up relatively quickly. If the heatsink doesn't come off quickly, try to wedge your tool in there as best you can to get some micro cracks going inside the epoxy, then come back to it again after another deep freeze.

Go slowly. And be aware when the card has reached room temperature so that you know when to stop. You don't want to over stress the solder balls on the underside of the chip.

Let us know what kind of product you decide to use for the heatsink and thermal epoxy.

TrashPanda wrote on 2022-06-17, 19:37:

Youll have to fight the P3 Tually over it .. it runs circles around the P4 ...even with a GeForce 3 ti500 the P3 handily beats it, Intel really did shit the bed with Netburst and its super long pipeline. The only case where the P4 manages to show its strength is when you run software than can saturate the DDR bandwidth on the P3, the P4 simply has more but for gaming . .its no contest the P3 beats it.

My argument is always that it was extremely successful for the time. One of the best indicators to me is that I could use it very successfully for for five years, and more after that as a secondary machine.

This is hard to grasp for people nowadays who didn't grow up with computers of the 80s and 90s. It was around this time that it was impossible to buy a computer that wasn't going to become grossly outdated within about 6 months. 5 years is a great run, for any computer.

The Pentium 3 was a great CPU. But I think that the first processor with staying power was the Pentium 4, at any speed. The next slowest was the 1.3Ghz, so I was already near the bottom of the barrel.

Kahenraz wrote on 2022-06-17, 20:50:

I used my freezer and a nylon spudger to remove the absolutely cemented-on heatsink of a Radeon 8500 (don't ever use a screwdriv […]
Show full quote

Radical Vision wrote on 2022-06-17, 19:36:

So as i found that the temperatures on the Voodoo V 5500 chips are dumb high (specially the primary chip closest to the VGA port is really burning fingers....) and i never like to use hardware that cook itself slowly over the years (sure these cards working fine like that, but still they may die at some point from overheating..), so i plan to freeze the card in freezer and then to pop off the garbage joke of a heatsinks and replace them with at least 5 times bigger heatsinks and fans, or maybe even Zalman coolers.

I used my freezer and a nylon spudger to remove the absolutely cemented-on heatsink of a Radeon 8500 (don't ever use a screwdriver). It took two or three freezes until it finally broke off with a satisfying crack. Graphics cards have a lot of copper, and although it takes a long time to cool, they will heat up relatively quickly. If the heatsink doesn't come off quickly, try to wedge your tool in there as best you can to get some micro cracks going inside the epoxy, then come back to it again after another deep freeze.

Go slowly. And be aware when the card has reached room temperature so that you know when to stop. You don't want to over stress the solder balls on the underside of the chip.

Let us know what kind of product you decide to use for the heatsink and thermal epoxy.

I see.. Well i dont have nylon spudger, but i will look arround i may find something useful.. There is a tool for bikes that is used to add/ remove the tire more easy and is from plastic/ nylon so that may come in handy. But i think i will try similar method as the guy from the video... Guess putting the Voodoo V in the freezer for whole nigh should to the trick after all it will get it below subzero temperature ( i think the freezer on the fridge was set to -6c that should be more then enough) and will prob work fine... Good advice on doing the whole thing fast, and not giving time to the card to warm up so the thermal cement too..

I think to use a plastic cover for the card as the guy did and a knife or screwdriver but only on the heatsink ofc (after all that is what for he put the platic card to protect the PCB from the screwdriver), i hope for the best, as currently the main chip is running way over 60c under idle that is BS... But after all these heatsinks 3DFx used to put on the cards are just awful pieces of dog shit... They could at least use the same heatdsinks as they did on Voodoo III 2K cards, as every time i try to cool Voodoo III 2K only a fan is needed, while Voodoo V sucks it, guess there was something they did change that made the chips even more hot..

Also improving the cooling will allow for OC of the chips, so instead of 166MHz i may end up with 188MHz or even 200MHz... But need to think about that, as i may end up messing up the card, so will see OC of 188MHz sounds fine, as going more then that is not very good idea... I cant on other hand OC the Pentiums, as it seems they get UNstable on 1.5GHz as for some reason the AOpen fails at setting the voltage i want, i set some value and every time post screen reports stock voltage wtf...

This is the tool I'm talking about. I use these for scraping, prying, spudging; anything where I would rather damage the tool than the thing I'm working on. The nylon is still hard, but much softer than most plastics, like ABS. They come in all kinds of colors. My favorite is the orange, since it's easy to find again after I set it down.

I have a lot of these. They're cheap, and can be treated as disposable. Cut them in half, sharpen them on a stone, make a fine point with a knife. They are extremely versatile. They are also excellent at scraping off expoxy without the risk of damaging a die.

The problem with wedging in a screwdriver is that it's much harder than the GPU. If you apply too much pressure, you'll crack the die or the substrate. Using a screwdriver is never a good idea for this.

For removing expoxy, the three key solvents are: citrus (Goo Gone), 70% isopropyl alcohol, and acetone. Use acetone only on metal or other surface when it's otherwise impossible, as it is an extremely potent solved. Most thermal paste will come off with citrus, but some epoxies do need acetone.

Just a tip - instead of freezer - use instant freeze spray (ie. something like that https://www.nteinc.com/chemtronics/freeze-spray.php ). Liquid nitrogen does the work as well as long as you have access to it. Also thermal stress can cause damage to bga solder joints, and I've read somewhere it will possibly shorten the life of the caps.

But to be honest - IMO there is not much sense in using rare parts with a lot of sentimental or monetary value on a daily basis - so better to keep them safe in original condition and occassional use won't do any harm. There is a lot of other cards/CPUs/motherboards which are much more common and will do the job just fine, and if they die - you just can put another one in it's place without shedding a tear...

HanJammer wrote on 2022-06-17, 22:15:

Just a tip - instead of freezer - use instant freeze spray (ie. something like that https://www.nteinc.com/chemtronics/freeze-spray.php ). Liquid nitrogen does the work as well as long as you have access to it. You don't want to put entire card in the freezer unless you want to shorten the life of the caps significantly.

You can use a can of compressed air, turn it upside down and spray the propellant onto the heatsink till it freezes, its a cheap way of doing this.

Just do it in an open well ventilated place so you dont gas yourself.

TrashPanda wrote on 2022-06-17, 22:19:

You can use a can of compressed air, turn it upside down and spray the propellant onto the heatsink till it freezes, its a cheap way of doing this.

You can but it can leave residue (while dedicated freeze spray by design should do that) and is not as convenient to use. Also it frequently is flammable.

I think that the freezer is much more gentle than freeze spray. It's not the temperature difference as much as it is the shock of rapidly moving from one temperature gradient to another. The act of freezing and then wrenching the heatsink off applies kinetic stress at a very specific place. Rapid freezing has a much higher chance of causing cracked solder joints since the temperature change happens rapidly and unevenly.

This is what a sponge is used for when you want to clean the tip of your soldering iron. You're not wiping as much as thermally shocking the top to contract suddenly and the solder falls off. The principal is the same with joints. A very sudden change in temperature (thermal shock) is the absolute harshest thing that you can subject any kind of metal to. The principle is the same when molten metal touches water; they can explode.

Another example is when you rapidly cool an ice cube by putting it into warm water. It cracks. But warm it slowly, and it won't.

Kahenraz wrote on 2022-06-17, 19:22:

Fun fact. My parents bought me a computer around the year 2000 that was supposed to "last me through college", and it was a Pent […]
Show full quote

TrashPanda wrote on 2022-06-17, 12:54:

Kahenraz wrote on 2022-06-17, 11:44:

That MX 440 with a 32-bit bus sounds wonderfully terrible. I want one.

Sounds perfect for my Rambus 1.4 P4

🤣

Fun fact. My parents bought me a computer around the year 2000 that was supposed to "last me through college", and it was a Pentium 4 1.4Ghz with RDRAM. So I am intimately familiar with this platform. 😀

I lasted about five years. It was a good investment, and really catapulted my technical prowess. I did a lot of work during that time with computers and technology and was able to save enough money to pay for the next upgrade myself. In in a sense, it did last me through college.

The first problem I ran into was the 256MB of memory. It was more than enough for Windows 98. It was also fine in Windows 2000, albeit with more frequent disk swapping, but it was not enough for Windows XP, which really chugged. I wanted to upgrade the memory for some time, but RDRAM prices were stupid expensive. So I instead saved that money to put towards a whole new computer.

My next upgrade was an Athlon 64 3400+ with 3GB of RAM and a GeForce 68000 GT. The uplift was incredible.

I never really thought of my Pentium 4 as slow. In fact, for years, anything I threw at it ran very well. I had a 19" monitor and a GeForce 2 Ultra, and could run everything at 1600x1200 with max FPS for years. It was also an incredibly stable platform. I never had issues with crashes or instability from the hardware ever.

So, yeah. On paper, these processors aren't much to look at, and they aged very poorly as the Pentium 4 line progressed. But for the time, it was a great platform.

Also quite ironic, an MX 440 would be about as fast as my GeForce 2 Ultra. So you're basically making fun of the same configuration I grew up with in my Kate childhood. I still think it was a great system, and I'll fight you to defend it. Haha.

Well, I assume the RAMBUS did a significant uplift in performance.. I bought an old Fujitsu-Siemens P4 with 1.4 GHz. It came with SD-Ram, and its performance was a real PITA 😄

TrashPanda wrote on 2022-06-17, 22:19:

Just do it in an open well ventilated place so you dont gas yourself.

Or close all your windows and enjoy the high 😀

debs3759 wrote on 2022-06-17, 23:32:

TrashPanda wrote on 2022-06-17, 22:19:

Just do it in an open well ventilated place so you dont gas yourself.

Or close all your windows and enjoy the high 😀

Better not, some air dusters are pretty toxic and inhaled in large quantities can cause brain damage and neurological problems... 🙁
What happened to Gordo in "Halt and Catch Fire" is a serious threat to any guys doing some more serious work with electronics (yet I can't get used to using proper filters when soldering)...

HanJammer wrote on 2022-06-17, 23:36:

debs3759 wrote on 2022-06-17, 23:32:

TrashPanda wrote on 2022-06-17, 22:19:

Just do it in an open well ventilated place so you dont gas yourself.

Or close all your windows and enjoy the high 😀

Better not, some air dusters are pretty toxic and inhaled in large quantities can cause brain damage and neurological problems... 🙁
What happened to Gordo in "Halt and Catch Fire" is a serious threat to any guys doing some more serious work with electronics (yet I can't get used to using proper filters when soldering)...

I dont know how it is elsewhere in the world but the most common propellants here are Propane and CO2, with CO2 mainly used for Foodstuffs and Propane used for most other aerosols, air dusters are the exception as laws here made the manufacturers change what they were using so you cant get high from them. (They can still kill you but you sure as fuck wont be high doing it)

Propane propellants can actually be very toxic in an enclosed space. See here for a story about a boy who died from too much canned deodorant.

https://youtu.be/RpZKdnOvNBo

Kahenraz wrote on 2022-06-18, 07:21:

Propane propellants can actually be very toxic in an enclosed space. See here for a story about a boy who died from too much canned deodorant.

https://youtu.be/RpZKdnOvNBo

mhm, why I suggested doing it in an open well ventilated space, even better if its a little windy.

As for why propane . .well Australia has one of the worlds largest natural gas reserves so propane is exceptionally cheap and plentiful here and education about not doing stupid shit with aerosol cans well known by early childhood. But kids will be kids especially indestructible teenagers and they think they know best, all we can do is try our best to steer them away from stupid destructive behavior and be there for them when inevitably shit goes down and they fuck it up.

Cotton wooling kids does nothing to help them learn about stupid behavior and the consequences of it.

BitWrangler wrote on 2022-06-17, 13:16:

Tetrium wrote on 2022-06-17, 09:53:

I was for instance surprised the HD5850 was supposedly a bit faster than the HD6850 even though the HD6850 was (when assuming from the name) supposed to be the HD5850's successor.

Yah the HD6950 was the more directly equivalent successor. There's a few things though where the HD6850 wins slightly over the 5850, but other things where 5850 makes it look quite bad. Due to better/longer support and I guess minor seeming tweaks, the 6850 stayed useful a lot longer. One or two "modern" games could be played on it even at 1024x768 or 720p. Approx equivalence is a GT650 non Ti. It's more like a 120W card than a 170W approx too, and depending on implementation you might run it with a single 6 pin PCIe power, though versions intended to appeal to overclockers had 2 like the 6870. Anyhoo, just saying a HD6850 is worth consideration on it's own merits, wide span of gaming years covered, lower stressed than the topper top end cards and a bit less heaty.

The die shrink is definitely nice. I always liked my HD6850 🙂. I got mine second hand to replace a HD5670 at some point and it lasted me years till that rig died.

Gmlb256 wrote on 2022-06-17, 17:28:

Radical Vision wrote on 2022-06-17, 16:33:

BitWrangler wrote on 2022-06-17, 14:10:

A moment of silence plzkthx, this weeks "one that got away"....

Intel branded Memory wtF is this...

Well, Intel was originally a memory manufacturer prior becoming involved in the CPU business. 😀

Intel branded SIMMs, that's neat! Learned something new today, unless I did already know and later forgot 🤣