Comparing The Power/Performance Of A NetBurst Celeron & Pentium 4 To Broadwell's Core i7 5775C

[Passso]

Yes 1 core VS 1 core would be interresting too.

Anyway this test shows how performances exploded in a few years even with the frequencies locked, that's incredible.

Here comes the second step, how will performances increase with both frequencies and process technology (nm) locked...

[rice_nine]

I wish there was a time axis on the Power Consumption graph so it was clear exactly how much sooner these tasks were completed. You know, in some places you can get shot for not labelling your axes... anyway, for a nice feature to consider adding, integrate and show the energy cost of the job and maybe even convert to $USD at some reference price like $0.15 * kW/hr for us lazy Americans. If this or something like this is already in PTS then never mind, I don't know because I don't mess with it.

P.s. maybe next time, boot the modern chip with maxcpus=1 so you're not benching a core against a compact cluster?

[Luke]

Actually quite an interesting test, because there are a HELL of a lot of Pentium 4 class machines still in daily use. In many communities people only have hand me down systems, hell it was not until 2008 and later that I saw Pentium 4's hit the dumpsters (for salvage) and until 2010 I still saw rooms full of Pentium 3's used for community Internet access. Now those are Pentium 4's, usually Prescott or similar. BTW, I back to back compared Prescott 3GHZ to an AMD Athlon 2.3 GHZ, for video playback at 540x960 the AMD just barely would do the job using Firefox as the player to simulate online playback. The P4 2.93 GHZ with hyperthreading had about 15% CPU to spare and used both threads. Of course, when Prescott came out, few applications could use hyperthreading, that has REALLY changed. Both machines will play 1080p with ease in MPV, but that's because both have modern graphics retrofitted and use VDPAU. I don't have any AGP cards from either AMD or Nvidia that can decode H264 in hardware, or the Pentium 4 Northwood 2GHZ would also be able to play 1080p in MPV. Using CPU decoding and XV output, both machines will handle 720p 30fps, neither will do 1080p w/o GPU decode. For a community center to replace ten computers would cost at least $2-3K, money that often simply is not available. Only the very newest would actually make a significant savings in electric bills-UNLESS the old ones are any kind of AMD K-7 generation (Athon XP or older but newer than K6). The K7 ran at full power dissipation at all times on every distro I ever ran one on, even with the CPU at idle. If a version of Firefox would come out that could use vdpau for H264 decoding in gstreamer, this would remove online video as the limiting factor in use of old or low powered machines as web browsers. The requrement would simply be anything with a PCI-e slot, or posession of one of the few AGP cards with H264 hardware decoding. The remaining issue would be the huge and heavy JS some websites use these days. Some sites will bring an Intel Atom N455 netbook to a crawl, these are about equal to a 2 GHZ non hyperthreaded Pentium 4 for both video playback and web browsing at 1/10th the power. For video rendering/H264 encode, the Pentium 4 or AMD Athlon will blow a netbook into the weeds, seen that many times.

[bison]

It would be much more interesting to see how a new Celeron compares to older high-end CPUs.

[stiiixy]

I just blew 10 bucks on an old (actually, it's new) ATI Radeon 7000 Mach64 and another 512MB SDRAM to add to me ol' Pentium III 1200MHz. A couple of HDD's also await testing....probably shoulda bought a PCI-SATA-port card.

[s_j_newbury]

It was just done at the defaults at the time, so that's how the testing remained today with being the reproducibility. Besides for most out there that's how the performance is/was aside from the source distributions or those rebuilding all their packages. Even if the P4 performance were still to double, as schmidtbag pointed out, it wouldn't gain much ground against the i7-5775C and the findings of this testing. For the march=native optimizations on x86_64 with the Broadwell system its performance too would have went up.

But that's just it; make it apples to apples, irrespective of the number of cores as others have mentioned which of course makes a huge difference to parallel code:

GCC from the NetBurst era optimized for contemporary CPUs, ie NetBurst by default. Modern GCC optimizes for contemporary CPUs with -mtune=generic, ie *not* NetBurst, the contemporary "generic" optimizations probably make the code slower! (Yes, NetBurst really was that bad with non-optimal code; avoid pipeline stalls at all costs!)

All that said, NetBurst was always a disaster IMHO, but it was faster at the time than represented here; or at least it was on the Gentoo systems I built! ;-)

[Gusar]

no offense but I do agree with the others, that this test - a multicore with hardware optimisations against a crusty old piece of ... well you know... a single core.

Well, I think the test as done already has merit. It does show how far we've come, especially in terms of power efficiency (schmidtbag makes a nice point about that). And most distros do not provide per-cpu optimized binaries, but generic ones. Not to mention that x264 (and probably flac too) has hand-written assembly for all the hot paths, so compiler flags won't make much difference. So I have no problems with the test as such, it's already interesting. I just think using a single thread would provide a nice addition.

[jakubo]

no offense but I do agree with the others, that this test - a multicore with hardware optimisations against a crusty old piece of ... well you know... a single core. Some sort of gentoo setup with native CPU optimisation and reduced core count could have told something about IPC, efficiency, handling of context switches or whatever. even quantitatively to some degree... in this case you can only say: look its faster! as expected. (hooray!)
I mean seriously, there will be exciting tests ahead, fiji, new Mesa capabilities, maybe some compiler tests - i dont recall seeing any in quite a while. will HSA affect the kernel scheduler by the way? maybe this is stupid but i really have no clue. It would only sound natural to me... Or simply take a day off. clean your head. focus on what is really important. Really no offense but reading this article was like... "Oh look, hes got quite some time to spare"
otoh if you wanted clicks... you certainly got them

[Michael]

I would like to se a x86 vs x86_64 on old hardware, somthing like P4 or Core 2 Duo.

I've been running such comparisons routinely for years, going back a ways I think I recall some Core 2 Duo or Quad tests.

[AJenbo]

I would like to se a x86 vs x86_64 on old hardware, somthing like P4 or Core 2 Duo.