Pentium D was it really that bad? Pentium 4 HT vs Pentium D + AMD Athlon x2 - Dual Core (UPDATED) 

Neither does "the D was smoother." :P Though in all seriousness, I'm sure it was smoother. Back in the Vista days, I upgraded from an A64 3000+ to an X2-3800+. The dual-core really did add a layer of creamy smoothness to the heavy new OS.

Agreed!

amd fx forever!

I wasted so damn much money on Pentium 4 refused to intel until 9th gen i5, lol.

I used to have a Northwood based (early) Pentium 4 system, that worked really well. Mostly played UT2003/2004 at that time. You have to understand back than we had really slow hard drives, that meant long loading times as well. Later I got a RAID0 Raptor setup, but I think with this one, I was still on a 5400 rpm HDD. That is something these retro test videos don't show. Also I keep on using Windows 2000 pretty long, because it had fewer background tasks runnig. And you would not just TAB out of a game, to switch your music. Single core CPUs had a really hard time with such actions. [EDIT] I also tended to have more memory than most builds, that allowed me to disable HDD swap. I don't want to downvalue this test - it is very true. But at the time, the whole system was much slower and other bottlenecks existed at the same time. [EDIT] If you test a retro setup with a modern SSD, you work around many issues of the time.

except that AMD FXes are having good days with their 13 years

It does perform better though. I had a 631 P4 in my HP DC7600, running Linux but no power to play RU-vid videos. Upgraded to a 945, and it manages to play 360P video’s without stuttering.

Thank you! This one meh..i just used old footage and filmed newer, but the scores were so hard to read in the old one.

@@jims_junk Can you try to add Fallout 3 or New vegas on your gaming tests in the future using these older cpus. Or even like far cry 1

@@marshal7969 Sure! Thanks for the suggestion. I'll see what I can do. I do need to add stuff

Yeah you are right, but do not forget that many software and games that we used did not support multi-thread anyways. The single core thread performance of an e8400 vs q6600 is impressive to say the least and faster then the 'slower' q6600 cores. I did my homelabs for MCSA with vm's on a Dell clampshell like this one with H.T. ! ;-) I remember it ran well and had no issue at all.

Oh man you're so lucky that you decided not to. We had school districts filled with them. They were awful! At one district we had one guy an intern who's job it was to just go around and replace hard drives and ghost the machine. At least a dozen a day died. There was no ventilation.

​@@jims_junkHow much of the massacre do you think was caused by the students?

HT was one way to squeeze some more performance out of the mediocre core with weak IPC. Netburst's pipeline was so deep that hyperthreading made perfect sense. The penalty of a miscalculation was just far too great! It's strange that Intel is now considering dumping HT altogether in favor of 'rentable' tile i.e a single tile can act like either a fast P-core or weak dual E-cores, depending on the workload. Now, on paper it, it sounds better than HT but in practice, who knows?!

@@Sam-K I still despise the P and E core garbage; very few things can make proper use of it. I won't use anything Intel newer than 11th generation.

Thank you! I really appreciate that.

RU-vid doesn't push his stuff I think because they're such short videos, I think. That's what I like about em though, straight to the point. I try to promote the channel wherever I can without spamming.

As long as you don't plan to get any use out of it, its actually fun to mess with something so bad. Just make sure you turn on your A/C first. Or heat an entire room in the winter. In my previous life working in IT, we had this one client that I talked about in the video. One PC the user complained was locking up. I took off the side and found that the intel heatsink had melted its plastic mounts and was just dangling from the fan wire. It was extremely clogged with dust, but still it got so hot that it melted the plastic that held it to the board.

@@jims_junk credit where due though, the thermal protection works really well! 😀

@@Aruneh LOL YEAH!

Core2 had shared caches... Besides that aspect of architecture making Pentium D terrible is just an urban myth. If that was an issue then preventing e.g. game (cause games were most susceptible to memory/cache performance) from being shuffled between cores to stick to single core (note: games ran single threaded back then) would make big difference in performance. Yet nothing like that happened - also why you always hear how bad the design was but never get any benchmarks to show how it affected users. Instead it was Athlon 64 X2 users had to change affinity of games (and some software) because cores in X2 ran their times with different tick values causing software reading these values to crap out when being moved from core to core. AMD fixed this bug in their K10 architecture... and maybe even later revisions of K8 X2 and why I think it was a bug.

​@@e8root Core2Quad has two dies with two dual core chips, each die has it's own L2 cache. Penryn is a good example, 12mb L2 combined but each dual core sees it as 6mb of L2. Half the die is also said L2 cache interestingly.

@@TheLawnWanderer C2Q sounds like Pentium D indeed, just with dual-core modules. Though if there was any communication mechanism between cores or it was just FSB like Pentium D I don't know. Intel could cook up something there to help with the performance. Either way just like with Pentium D it was not widely known you can get some performance in games when making sure game only uses one dual-core block and e.g. other programs can only run on second one. In fact since performance of C2Q and C2D in games with the same clock was nearly identical I don't think there was much to performance to squeeze. Same with Pentium D.

stock heatsink or larger non intel heatsink?

@@jims_junk I have also experienced that the temperatures (especially the latest D0 revisions) had absolutely no issue. More precisely, I had one Pentium 4 631 sSpec SL9KG and goes to idle below 40 with EIST/C1E on etc.. If I remember corectly, it was accompanied by a "large" standard Intel stock cooler (not the low-profile ones that Intel then sold with later C2D), but it didn't have a copper base. The problem was rather that the specification and solution were wrong. Many coolers did not fit very well. Those plastics legs simply broke and litterally it was one-time instalation product. And if, IF! was installed correctly, it works well.

lol and I've had people comment on the original video saying that 'there's no way it runs that hot'. Some don't realize what great space heaters they were.

probably ran at a lower resolution and/or lower quality settings. I kept the settings kind of high. Not trying to get the best score, but just keeping it even across the tests while pushing the cpu hard.

Exactly

No. Chiplets are substrate interconnect tech, while pentium D was just 2 dies separated

Had*? Or do you imply that modern ryzen are fake?

@@Перчатки I mean AMD called P4D and C2 fake dual and quads, so Ryzen/TR are fakes.

What are you blathering about? Glue logic is the circuitry that connects two disparate ICs, it's not a slur. Pentium D was an utter disaster. Bulldozer is a similar-scale disaster. Ryzen is leading Intel in... quite literally everything from gaming to servers for not only performance per watt, but also cores per socket and outright performance period. Equating Infinity Fabric to forcing all inter-core communication through the FSB is the silliest thing I've heard in a while.

@@tim3172 What are you even blathering about? AMD called Intel for gluing their chips back in P4D and early C2D and C2Q. Except, Aymdiots still claim that Faildozer wasn't a fail, and it's the only design from AMD in-house and not relying on Jim Keller. Speaking of Ryzen, it's merely a souped up K10 uarchitecture, modularised and modernised. This shows that AMD has no good engineers and had to rely on Jim Keller. Except, AMD went full Intel by releasing a glorfied low-end 6c for US$350-400!!! That's insane! AMD continues to sell a glorified low-end 6c to this day! That's insane as well! Plus, I remember AMD calling out Intel for being greedy by making 2S2P (2 sockets, 2 platform), guess what AMD pulled? Because it is and advanced version of FSB, since IF rely on memory speed and its latency, which was gimped in Zen++, and further gimped in Zen++++. Oh, it's also has latency. What? Not allowing the IF-MC to run 1:1 when it reached the thresold is worst than 2:1? If you must know, it's inversely proportional, basic mathematics. If the memory is running 2GHz (using 4GT/s kit), the IF is running at 2GHz at that latency. Now with the gimped Zen++, the IF is running at 1GHz.