To make the benchmarks more meaningful it would be good to include the results of the AMD processor this theoretical Pentium D would compete against.

alexanrs wrote:

To make the benchmarks more meaningful it would be good to include the results of the AMD processor this theoretical Pentium D would compete against.

It would and I will 😀.

The Windows 7 benchmarks scores are mostly to compare Windows 7 scores with the XP scores which will be added later (now done) (now moved).
When I have benched this system in Windows XP I will build an AM2 system to see if I can beat this systems scores (now done).

Here are some more benchmarks with other CPUs from year 2005-2006 to compare with.

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentum D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

Pentium D 995

Athlon X2 6000+

Conroe 1M E2140@3.0

I laughed so hard at the title reference 🤣 . I like the idea, it is very VOGONSy and I will await for further tests, perhaps with some early Core 2 Duos as well? I too want to laugh at the Pentium D 😜.

One note though, had this actually happened, Intel would have sold slower, hotter hardware, but at a high price and people would still buy it. 😵

F2bnp wrote:

I laughed so hard at the title reference 🤣 . I like the idea, it is very VOGONSy and I will await for further tests, perhaps with some early Core 2 Duos as well? I too want to laugh at the Pentium D 😜.

One note though, had this actually happened, Intel would have sold slower, hotter hardware, but at a high price and people would still buy it. 😵

I have just done some benchmarking in Windows XP and so far the Pentium D is not as slow as I would have liked, I think the Athlon 64 X2 AM2 system will have to be on the high end of the scale to beat this system. The AMD system will not run as hot or use as much power thats for certain.

I just realized that the Pentium D 950 isnt multiplier locked downwards.

I knew I could use the 12x multiplier as that is the multiplier the CPU idles at (Speedstep) but it seems I can also use what ever multiplier between 12x and 17x. I will need to run some more tests as this opens up more possabilities. The 12X CPU multiplier wasnt that interesting because I can only use the 4x memory multiplier (auto). If I use any other memory multiplier or chipset straps than the auto settings the system wont overclock. The 15x and 16x CPU multipliers are however very interesting because a slightly higher FSB will take the memory twords the limits for what is possible with DDR2.

If I will have to overclock the AM2 system to beat this system its only fair that I have tested the fantasy Pentium D 995 at tweaked settings.
My early C1 stepping wont go much higher than 4533MHz (4800 needs 1.65V) with reasonable voltage on air cooling but the FSB has a large impact on performance aswell.

Does someone know if all Pentium D CPUs have unlocked multipliers downwards like Core 2 Duo and later Intel CPUs?

Surprisingly the Pentium D did not gain much from increased FSB, without HT there isnt much need I guess.

I ran all the tests at 15*303 MHz which produces about the same CPU speed as I used above
The only difference in settings were that I had to change tRCD from 4 to 5 and tRAS from 13 to 14.
DDR2 @ 1212 MHz running CL4 is some kind of personal record and with only 2.25V! 😀

The benchmark scores were about the same as before. Well the Everest memory and some CPU benchmarks gained quite alot but 3dmark01/03/06 all gained less than 1%.
No use in posting screenshots with the same scores once more.

I did also run the tests at 4800 MHz (16*300) with the memory at 1200 MHz to see how an overclocked "Pentium D 985" would perform.
I will post some results in the "3dmark 2001" thread, "SuperPi thread" and the "Doom 3" thread for now.

The fastest AMD CPU released within a year from the Core 2 Duo was the Athlon X2 6000+.
I will now build a system around this CPU to see if it can beat the "Pentium D 995".

Remember than pre Nehalem Intel CPUs don't have a memory controller built in but rather it's in the northbridge and it has its own clock/timing/voltage settings. The clock seems to be whatever you set the fsb* to and timings are indirectly set based on the strap? There is one northbridge timing that you can set in your bios called tRD and it can have a dramatic impact on overall memory performance if not tightened http://www.anandtech.com/show/2427/7. In your bios it can be changed with the Static tRead Value setting and also I think the Performance Enhance setting will change it. You can verify (and change!) its setting in windows with the memset program's "Performance Level" reading. http://www.tweakers.fr/memset.html website seems to be down now but it shouldn't be difficult to find a download mirror.

On my p35 based motherboard the bootable values for tRD were different based on what FSB and memory ratio I was using but were generally in the range of 3-7 at very low FSB or 7-15 at high FSB, even though it lets me set it higher (slower) it wouldn't boot.

*Its more useful for comparison to express the tRD limit in time(ns) rather than clocks that are always going to be different lengths but couldn't find much info on the northbridge clock speed. I used a memory latency test that give a value in nanoseconds and used that to calculate the clockspeed from the difference of setting only the tRD to different values. I found it generally tracked the FSB +-4% but on the 5:8 memory divider it was about +15% and on the 2:3 it was -10% so I don't fully understand that behavior yet. Also someone mentioned that changing the cpu multi in relation to its default multi is what determines the northbridge clock on the 945 or 965 chipset but I can't seem to find that post anymore and I've only been testing with my e8400 underclocked with its 6x multi so far but I'll get around to checking the other multi's eventually.

Thanks for the information.

I know the P35/X38/X48 platform very well though 😀

Im running a very tight tRD of 2 at [FSB 267 mem 1066], [FSB 333 mem 1200] and [FSB 303 mem 1212]. It works well with stock MCH volltage at 1066 and with +0.2V MCH voltage at 1200 and 1212. The divider is 4x but set to auto.

It is possible to use all straps if I manually set all timings including tRD as the board dosnt do this correctly with this CPU for whatever reason. I think the BIOS is optimized for 333 Mhz and 400 Mhz FSB. The thing is that I have tried every strap with correct not so tight and tight tRD values and the memory just wont clock half as good as if I set both the memory divider option and strap option to auto (Dividers and straps are acually the same option in Gigabytes BIOS, dividers with letters marking straps so you cant set one without the other.), and then manually type in all the timings. This holds true when clocking memory with my E8600 aswell.

Perhaps Gigabyte had all their BIOS programmers optimizing the BIOS for the X48-DQ6 while the intern made the BIOS for the X48-DS4, I am considering a crossflash.

https://www.youtube.com/watch?v=ERmdTGJxIhg

Im using this post to dump links to images removed from the posts above to make room for other pics

______________________________________________________________

Windows 7 experience index: This is with the 7900 GTX, the GTX 285 got higher video scores (7.3 + 7.3)
http://u.cubeupload.com/jonaz81a/f1fPentiumD9504533X48DS.jpg

Windows 7 64 processes: When running Windows 7 64 with only 2GB memory its a good idea to get rid of unnecessary fat.
http://u.cubeupload.com/jonaz81a/e8bPentiumD9504533X48DS.jpg

Windows XP Processes: Not really needed to trim fat in XP with 2GB memory but I do it out of habit.
http://u.cubeupload.com/jonaz81a/840PentiumD9504533Gigab.jpg

Windows 7 64 memory usage: With less fat 2 GB memory is plenty.
http://u.cubeupload.com/jonaz81a/87cPentiumD9504533X48DS.jpg

Windows XP memory usage: Or lack of memory usage.
http://u.cubeupload.com/jonaz81a/f5dPentiumD9504533Gigab.jpg

1080P Bluray decoding: Pentium D 995 Win 7 64 with MPC HC-64 : The CPU is used for decoding.
http://u.cubeupload.com/jonaz81a/408PentiumD9504533X48DS.jpg

1080P Bluray decoding: Pentium D 995, Win XP with MPC-HC-32, The CPU is úsed for decoding (screenshot bugged)
http://u.cubeupload.com/jonaz81a/116PentiumD9504533Gigab.jpg

1080P Bluray decoding athlon X2 6000+, Win XP with MPC_HC-32, The CPU is used for decoding. (screenshot bugged)
http://u.cubeupload.com/jonaz81a/234A646000plusM2NSLIDel.jpg

SuperPi 1M: Pentium D 995, Win 7 64, , I forgot to run this test at 4533 MHz so this score is at a slightly higher frequency but with loser memory timings.
http://u.cubeupload.com/jonaz81a/PentiumD9504675X48DS.jpg

Youtube 720P (Flash): Pentium D 995, Win 7 64, The CPU is used for decoding.
http://u.cubeupload.com/jonaz81a/d92PentiumD9504533X48DS.jpg

Youtube 720P (Flash): Pentium D 995, Win XP, The CPU is used for decoding.
http://u.cubeupload.com/jonaz81a/546PentiumD9504533Gigab.jpg

Youtube 720P (Flash): Athlon X2 6000+, Win XP, The CPU is used for decoding.
http://u.cubeupload.com/jonaz81a/5A646000plusM2NSLIDel.jpg

Youtube 1080P (Flash): Pentium D 995, Win 7 64, The CPU is used for decoding and its actually smooth but just barely
http://u.cubeupload.com/jonaz81a/5d9PentiumD9504533X48DS.jpg

3dmark 2001: Pentium D 995, Win 7 64, This is with the Geforce GTX 285, the score is crappy as always in Windows Vista and Windows 7/8
http://u.cubeupload.com/jonaz81a/PentiumD9504533X48DS.jpg

3dmark 2006: Pentium D 995, Win 7 64, Using the GTX 285 the system gets a decent score.
http://u.cubeupload.com/jonaz81a/31ePentiumD9504533X48DS.jpg

3DMark 2001: Pentium D 995,Geforce 7900 GTX, Windows 7 with the latest driver for the 7900 GTX.
http://u.cubeupload.com/jonaz81a/361PentiumD9504533X48DS.jpg

3DMark 2001: Pentium D 995, Geforce 7900 GTX, Windows XP with driver 92.91. Almost double the score...
http://u.cubeupload.com/jonaz81a/PentiumD9504533Gigab.jpg

3dmark 2001: Athlon X2 6000+, Geforce 7900 GTX, Windows XP with driver 92.91
http://u.cubeupload.com/jonaz81a/A646000plusM2NSLIDel.jpg

3dmark 2003: Pentium D 995, 7900 GTX, Windows XP with driver 92.91
http://u.cubeupload.com/jonaz81a/a33PentiumD9504533Gigab.jpg

3dmark 2003: Athlon X2 6000+, 7900 GTX, Windows XP with driver 92.91
http://u.cubeupload.com/jonaz81a/6c4A646000plusM2NSLIDel.jpg

3Dmark 06: Pentium D 995, Windows XP, 7900 GTX with driver 92.91
http://u.cubeupload.com/jonaz81a/fcePentiumD9504533Gigab.jpg

3Dmark 06: Athlon X2 6000+, Windows XP, 7900 GTX with driver 92.91
http://u.cubeupload.com/jonaz81a/14aA646000plusM2NSLIDel.jpg

Doom 3 1024*768 Ultra: Pentium D 995, Windows XP, 7900 GTX, with driver 92.91
http://u.cubeupload.com/jonaz81a/525PentiumD9504533Gigab.jpg

Doom 3 1024*768 Ultra: Athlon X2 6000+, Windows XP, 7900 GTX, with driver 92.91
http://u.cubeupload.com/jonaz81a/ed7A646000plusM2NSLIDel.jpg

_____________________________________________________________

[Windows 7 with the Pentium D 995]

I tried Doom 3 at 1024*768 with ultra settings and got 138 FPS with the GTX 285 and 127.5 FPS with the 7900 GTX.

I tested both NFS Shift and Dirt 2 and I could max both games with extra everything at 1280*1024 with the GTX 285

I did also try Crysis (DX10, both 64bit and 32bit). Medium settings got ~75(50-100)FPS and high settings got ~45(30-60)FPS, both at 1280*1024

[/Windows 7 with the Pentium D 995]

All benchmarks updated.

The only thing I have left to do is overclock the Athlon X2 6000+ to see how it stacks up against the Pentium D @ 4.8.

Well I will also install some games among them Crysis but Im sure the Athlon X2 wont perform any worse than the Pentium D so the games will run fine.

Here is a short conclusion
________________________________________________________________________________________________

The hypothetical Pentium D 995 4533 Mhz CPU is about even in performance compared to an Athlon X2 6000+
If the Core 2 Duo would have been a year late Intel could just have used "Preslers" potential for high clockspeed.
The Athlons lower memory frequency may seem unfair but the Athlon runs CPC which makes alot of difference.

A stock Athlon X2 6000+ uses ~40W less power both at idle and load compared to the Pentium D at 4533 Mhz 1.45V.
The Athlon is the better CPU all in all but its sad that AMDs flagship mid 2007 still was a 0.09 micron Athlon 64 X2.
Just to put things in perspective, K8 launced mid 2003, 0.09 micron K8 a year later, the Athlon X2 a year after that.

________________________________________________________________________________________________

Perhaps I shoud build a system around a first generation Core 2 Duo CPU just to see the performance difference...

Laugh at Pentium D? Seriously? This level of performance is more than the average user typically uses NOW. Just because AMD "was teh better at teh times" does not mean Intel was crap.

That Pentium D would have been an absolute beast back then. Especially considering the X2 6000+ would have been pretty high end at the time. When I bought my Core 2 Duo E6400 it was ~$220 and it competed against X2 5000+', so the X2 6000+ was pretty strong for its time. Nice to see the Pentium D put up a good fight. Would be interesting to see the power usage.

Skyscraper wrote:

Surprisingly the Pentium D did not gain much from increased FSB, without HT there isnt much need I guess.

The Pentium EE *did* have HT however (which effectively made it the first 'quadcore' x86?).
So it would be interesting to know what would happen with a Pentium D at high clockspeeds and HT enabled.
Sadly you'd need a rare and expensive EE CPU for that in the first place.

My brother actually ran a Pentium D at about 4.2 GHz for a while.
It wasn't very impressive though. My Core2 Duo E6600 at 2.4 GHz could easily keep up most of the time.
And the E6600 clocked to 3 GHz (and 1333 FSB) with no effort at all, leaving the Pentium D (as well as any AMD offerings) way behind in its dust.

fyy wrote:

That Pentium D would have been an absolute beast back then. Especially considering the X2 6000+ would have been pretty high end at the time. When I bought my Core 2 Duo E6400 it was ~$220 and it competed against X2 5000+', so the X2 6000+ was pretty strong for its time. Nice to see the Pentium D put up a good fight. Would be interesting to see the power usage.

The power consumtion wasnt that bad, The Pentium D system peaked at ~315W at the wall when running at 4533 Mhz 1.45V with the Geforce GTX 285 running 3dmark 06.
The Athlon X2 system peaked at ~285W. The Athlon X2 6000+ has a TDP of 125W so it is also a power hungry CPU.

At least the overclocked Athlon X2 6000+ beats the Pentium D @ 4800 Mhz in most tests, but its not by much.
The E2160 "Conroe 1M" beats them both.

Doom 3 1024*768 Ultra, Pentium D 950 @ 4800: 157.9 FPS

Doom 3 1024*768 Ultra, Athlon X2 6000+ (0.09 micron) @ 3450: 162.4 FPS

Doom 3 1024*768 Ultra, Conroe 1M E2140@3.6: 187.9 FPS (100% overclock using air cooling!)

Skyscraper wrote:

The power consumtion wasnt that bad, The Pentium D system peaked at ~315W at the wall when running at 4533 Mhz 1.45V with the Geforce GTX 285 running 3dmark 06.
The Athlon X2 system peaked at ~285W. The Athlon X2 6000+ has a TDP of 125W so it is also a power hungry CPU.

That is a lot of power considering my core 2 duo + 7850 peaks 200W

Of course look at AMDs new CPUs. Space heater anyone?

smeezekitty wrote:

Skyscraper wrote:

The power consumtion wasnt that bad, The Pentium D system peaked at ~315W at the wall when running at 4533 Mhz 1.45V with the Geforce GTX 285 running 3dmark 06.
The Athlon X2 system peaked at ~285W. The Athlon X2 6000+ has a TDP of 125W so it is also a power hungry CPU.

That is a lot of power considering my core 2 duo + 7850 peaks 200W

Of course look at AMDs new CPUs. Space heater anyone?

I think the factory overclocked GTX 285 pulls some power aswell and it got pushed much harder than I would have thought with both CPUs.
The noise the GTX 285 makes during load is worth recording.

I would have liked to compare the Pentium D and the Athlon 6000+ with the Core 2 Duo E6700 as it was the Intel CPU at the same price point as the Athlon 6000+ the summer 2007.

My E6700 is in the mail from China and wont get here anytime soon so I thought that I would see how the above CPUs fare agianst Intels low end CPU at the time.

I bought a new Intel system the summer 2007 to replace a socket 939 gaming system. The stuff I bought was a Gigabyte P35-DS3 motherboard, an Intel Pentium Dual Core E2160 1.8 GHz CPU, 2GB 1066 MHz DDR2 memory and a cheap Radeon X1950 Pro. The E2160 was priced at $84 (dont rember what I paid in Sweden) and it was the smart choice compared to its even cheaper little brother the E2140 priced at $74 as the E2140 often got limited by its 8X multiplier when overclocking. I think it would be fun to see how this entry level CPU I soon replaced with an E6750 and later an E8600 will compare against the Pentium D and AMDs flagship the Athlon X2 6000.

I will not use my Gigabyte P35 DS3 or the X48-DS4 as I need to try out a MSI P35 Neo2 who has been collecting dust in the untested board pile.

Edit:

I just read this article at tomshardware, it seems there is a memory performance issue when overclocking with the MSI P35 board.
http://www.tomshardware.com/reviews/taking-e6 … -4ghz,1658.html

Anandtech got poor memory scores aswell.
http://www.anandtech.com/show/2392

Edit2:

Here is the issue found by xbitlabs
http://www.xbitlabs.com/articles/mainboards/d … platinum_8.html

It seems you can work around the issue so I will use the MSI board anyhow unless I find it underperforms unacceptably much.

Edit3:

Speaking of the devil, the E6700 arrived just when I said it would not be here for a while. Lucky enough Im done with the benching of the E2160.

I have removed alot of images to make room for the Conroe CPUs. The removed images are still linked in one of the posts above.
The benchmarks with the 7900 GTX only filled the purpose of highlighting GPU and CPU bottlenecks when compared to the GTX 285 benchmarks.
The Windows 7 benchmarks were mostly there to show how much faster Windows XP is 😀 .
The Bluray 1080P and Youtube 720P tests turned out to be useless as all tested CPUs can handle these tasks with ease.

I have decided to normalize all Conroe CPUs to 9*333 = 3 GHz. This is a speed they do using default voltage and they still have more "overclocking headroom" than the Athlon 6000+. The reason for this is that Intel was VERY conservative with the clocks on the Conroe CPUs and pretty much all "performance minded" people with tweakable motherboards did at least turn up the clocks a little bit while the Athlon X2 6000 @stock is very close to the 0.09 micron K8 clock ceiling. It is also nice to see the clock for clock difference between K8 and different Conroe versions.

Benchmarks updated.

When clocked to 3.0 GHz the E2160 "Pentium Dual Core" Conroe (Allendale-1M) CPU with only 1MB Cache memory beats both the Pentium D and Athlon 64 X2 in all 3D tests
At 3.6 GHz the E2160 is far ahead of the Athlon X2 at 3450 MHz and the Pentium D at 4800 MHz in all 3D tests and many other tests.

The E2160 + GTX285 used 215W peak power at 3.0 (100W less than the Pentium D@4533) and 256W peak power at 3.6 (29W less than the X2 6000+ at stock speed).
The AMD platform is different so lets say that is at least part of the explanation for the power usage.