Phenom II vs Core 2 Duo - Benchmarks

[massive88]

I decided I might as well make this its own thread, if anyone else has some head to head comparisons for different processor types, feel free to add it here as well, or if anyone has any specific games they wish for me to test, let me know and Ill run it.

I have two very similar computers, one is running a core2 Duo at 3.2 Ghz overclock, with 2 gig ram, and WinXP 64-bit.  My other computer is running an AMD Phenom II X3 at 3.19 Ghz overclock, with 2 gig ram, and Win7 64-bit.

The Core2 Duo machine is the better machine, it has faster ram, and a better video card.  The AMD machine is running on an old AM2 socket motherboard from 2007 (thank you AMD for making the first next-gen processor that I didnt need to buy a new MoBoard for!).  While the AMD machine has a third core, but I did not do anything special to the affinity to aid it.  All in all I think its a fairly good test.  The effects of OS differences I think are minimal (as evidenced by my previous tests of XP64 vs Vista64), and the hardware outside of processor should not have much of an effect either.  Certainly we all know that processor horsepower is the main driving force in emulation speed (mame in particular) so I think this represents a fairly good test.

I ran Mame 0.132, compiled freshly from source using Headkaze's Compiler, with the hiscore diff patch applied.  For the AMD I optomized for AMD 64, and for the Core 2 build I optimized for Core 2.  Both builds were done as well with 64-bit processor and Dual Core checked.

I ran two batteries of tests, one using 240 seconds of run time, and another using 440 seconds.  After looking at the data though, there is nothing additional provided by the additional, longer runs.  The results of the games, comparatively, was close enough to be considered the same (within +/- 2%, except for gauntleg, but I suspect theres more error in that one since both numbers jumped significantly, and thus is not a good test).  Here is the string used:

vmame.exe -noautoframeskip -frameskip 0 -seconds_to_run 240 -nothrottle -nosleep -video ddraw -skip_gameinfo -effect none -nowaitvsync -noread config -mt [ROMNAME]

Core2 Duo - 3.2 Ghz - 240s
blitz2k - 128.0%
gauntleg - 184.1%
gradius4 - 78.3%
kinst2 - 533.2%
radikalb - 97.2%
ridgerac - 103.3%
tekken3 - 149.3%
wargods - 275.8%

Phenom II - 3.19 Ghz - 240s
blitz2k - 135.7% +6%
gauntleg - 206.0% +12%
gradius4 - 100.6% +28%
kinst2 - 412.8% -23%
radikalb - 119.1% +23%
ridgerac - 101.3% -2%
tekken3 - 110.9% -26%
wargods - 224.1% -19%

So the conclusion remains, inconclusive.  Personally I'm surprised by the results, as I expected the Core2 to be faster across the board.  As with many things in Mame, it appears to be about the driver.  In 8 relatively randomly selected games, each processor was beaten badly in two, and barely in two, you cant ask for a much more standard randomness then that.  If anyone has the opportunity to double check the the difference between XP64 and Win7 64 as your OS Id appreciate it.  Though I'm fairly certain it wont make any difference, that is most likely the biggest margin for error in my tests.

As for "Bang for Your Buck", Newegg currently has X2's for $58 for 2.8 Ghz, up to $68 for 3.0 Ghz, and my X3 used in this test for $87 (2.9Ghz) while Core2 Duos are $145 for the 3.06 Ghz E7600, though mine tested was an E8200, and its equivalent at 3.0 Ghz is $168.  So for now, its safe to say imo, that the AMD X2 is the most efficient performance for your dollar, costing roughly a third of its Core2 counterpart, and besting it in some games.

Any thoughts or additional tests anyone wants to see?

[drventure]

Good stuff. Thanks for posting this!

I've been considering attempting an upgrade to a machine that can overclock to 4ghz, just to get at some of the 3d games that just won't run with my current AMD phenom x2, but overclocking just seems like so much black art to me.

Still, may need to do some more looking there.

[FrizzleFried]

The problem I see is that the core2duo's can be pushed to 3.7 with very very little effort...and 4ghz at least 50% of the time... how well do the AMD chips overclock?

[bkenobi]

With the poor cooling that we know exists inside a cab, why would you overstress things by overclocking?  It seems to me that the best course of action (for a cab installation) is to get the best bang for your buck (or the fastest if that's your choice) setup and leave it at stock speeds.  Since you have to upgrade cooling anyway, adding the extra heat from overclocking would be tough to deal with short of liquid cooling, right?   

[massive88]

The problem I see is that the core2duo's can be pushed to 3.7 with very very little effort...and 4ghz at least 50% of the time... how well do the AMD chips overclock?

Thats a good point and something I don't know for certain, you'd probably have to look online for the answers.  Id guess AMD should overclock well,  since its also a low power (65W) chip and small die (45nm) which are usually recipes for good overclocking.

On my machines I have my 2.6 Core2Duo clocked up to 3.2 Ghz with an aftermarket cooler easily (I was stable at 3.33 but backed it down).  The AMD is running on its stock cooler (Aluminum heatsink, small fan) right now, and clocked to 3.19 from 2.9 on its first try, by only changing the core Mhz, so I would expect the overhead is pretty good, but I dont know for sure and wont be pushing it any since its my main computer.  

At any rate, if the question is whats the fastest computer I can get, Id say go with the higher stock Core2 duos, but they come at a price premium.  If its more a question of value, then it appears the X2's are an excellent choice.  It would be hard to justify 2-3x the cost for 1.33x the performance (assuming 3ghz against 4 ghz) from a value perspective.  Though from what Ive seen on my chip, I would imagine getting 3.5Ghz at least out of it should be easily done, so you are talking about another 500mhz at most for that price difference.

With the poor cooling that we know exists inside a cab, why would you overstress things by overclocking?  It seems to me that the best course of action (for a cab installation) is to get the best bang for your buck (or the fastest if that's your choice) setup and leave it at stock speeds.  Since you have to upgrade cooling anyway, adding the extra heat from overclocking would be tough to deal with short of liquid cooling, right?   

Not everyone has the traditional closed up Cab design, you can always have a fan sucking in outside air directly as well.  At any rate, when talking performance I think overclocking should be considered, even if its not for everyone.  With the price difference being as extreme as it is between the AMD and Intel offerings, it seems to me to be a clear cut case of value or max performance, with the AMD crushing in the value department, at least as it seems on my tests.

[bkenobi]

...with the AMD crushing in the value department...

That's been AMD's MO forever.  They were a nobody when it came to power back in the day though (think 486 and early 586).  Then AMD actually made a good chip (the 686 line) and started to take market share.  Since they were cheaper and performed better, it was a no brainer to go AMD.  At this point, they don't always have a better chip but they almost always have the best bang for the buck.  I haven't run an Intel chip since I was running my P3 600 back in school.

To be fair though, Intel does have a value line that does compete with the AMD on price quite nicely.  Unfortunately, it usually lags on performance.

[FrizzleFried]

With the poor cooling that we know exists inside a cab, why would you overstress things by overclocking?  It seems to me that the best course of action (for a cab installation) is to get the best bang for your buck (or the fastest if that's your choice) setup and leave it at stock speeds.  Since you have to upgrade cooling anyway, adding the extra heat from overclocking would be tough to deal with short of liquid cooling, right?   

Been running my Core2Duo at 3.7ghz up from 3.00ghz since day one...going on about a year and a half now.  Plus,  who says cabs have to have poor ventilation?  All it takes is a couple 120mm fans.

[solid12345]

I have an e8400 core duo that my friend OC'ed to 4.05 GHZ, nothing but a basic fan running on top of it and been using it for a year and a half with no heating problems, just amazing.  I'd love to use it as a MAME machine as it can even run Gauntlet Legends pretty damn well but alas it is much more needed for my freelance design work.

[Erik]

Thanks for posting!  I was wondering about the Gauntlet Legends numbers (184%+ fps) Does it remain playable throughout or does it drop to low fps sometimes and disturb the gameplay?

[u_rebelscum]

Thanks!  Great to see numbers, side to side.  I didn't think PhenomII could split with a core2 at the same Ghz.  A couple thoughts...

I have two very similar computers, one is running a core2 Duo at 3.2 Ghz overclock, with 2 gig ram, and WinXP 64-bit.  My other computer is running an AMD Phenom II X3 at 3.19 Ghz overclock, with 2 gig ram, and Win7 64-bit...

Close, but not quite apple to apples.  It's great to see both running at same Ghz & memory, but two core vs three core.  I'm not sure how much difference this makes, but....  A few games in mame can use 3+ cores fairly well, while most games only use the first two well.  I wonder how those better threaded games match to 2 vs 3 core numbers.  (I did a quick search for a list of well threaded games, but no luck.)  Can you run these again with -nomt, to test which games can use the extra cores more?

I'm pretty sure the OS difference it's much, as you said, but I would how much?  0%?  5%?  +2% to -2%, depending on game?

I ran Mame 0.132, compiled freshly from source using Headkaze's Compiler, with the hiscore diff patch applied.  For the AMD I optomized for AMD 64, and for the Core 2 build I optimized for Core 2.  Both builds were done as well with 64-bit processor and Dual Core checked.

I wonder how mame 0.136, precompiled binary d/l from mamedev.org, compares.  (With -str, you don't have to worry about nag screens, and hiscore was removed because of the bugs it added to mame.)

I ran two batteries of tests, one using 240 seconds of run time, and another using 440 seconds.  After looking at the data though, there is nothing additional provided by the additional, longer runs.

Thanks for running the longer tests.  I've been suspisous of JohnIV's 90 second tests. 

Here is the string used:

vmame.exe -noautoframeskip -frameskip 0 -seconds_to_run 240 -nothrottle -nosleep -video ddraw -skip_gameinfo -effect none -nowaitvsync -noread config -mt [ROMNAME]

Why ddraw?  The default since 0.107 has been d3d, MS depreciated ddraw in dx8, and most cards are better at d3d now a days.

Than again, the mameUI benchmarks show little difference between -video none and -video d3d.  Is it the same with ddraw and your video cards?

Any thoughts or additional tests anyone wants to see?

This is just a summary of the requests I put throughout my post.  Don't worry about testing all games, and 120 or even 90 seconds might be good enough, but then the time wouldn't match your prior tests.

-nomt (to see which, if any, games run anything close to 2 or 3x faster, depending CPU)
-video d3d (to see if -ddraw is slowing anything down)
Official precompiled mame 0.136 binary (not optimized).

Lastly, can you try running dolphin?  It's still primary, and you'll need mame than 0.134 or newer.  Very big CPU hog.

Again, thanks for posting the numbers!

[massive88]

Ill try to get around to doing some of your suggestions this weekend Robin.

Im curious as to why you would care about the non-optimized versions?  The reason I went with the optimized, is that in a real world scenario, that would be the version a person would be using.  Imo it doesnt matter much how each processor does the precompiled, but how well it can run mame.

I didnt know that any games used anything beyond 2 cores.  Ill definitely rerun with MT off completely, that should give a fairer comparison between the two, probably should have done that in the first place.  I imagine the use of mt should be a fairly consistant percentage no matter which processor, though I guess I can see that as well when I get it run.

The -ddraw flag, and in fact the entire line, was just taken straight from the 4ghz thread.

When I compared Vista to XP a while back, I was getting numbers that were within their own tolerance.  IE you might get 410 421 415 running Xp, then Id run once in vista and get 412.  As Win7 is supposed to be faster than Vista, but probably not faster than XP, speaking in incredibly generic terms, I think theres probably not much error there.

Ill try to grab Dolphin, though its probably not in my current set since Im at 0.132...

Peabody:  Sorry I didnt test gauntlet actually playing it.  I would guess its slow as on my Core2 it still plays slow, and the difference between then in the benchmark is not much.  Theres a lot of FMV in the gauntlet attract screen, which probably skews the benchmarks.  You could probably get a better number playing with the duration to cut it off after the max in game running.

[Kman-Sweden]

Posted this in another thread earlier..

XP v.s Windows 7
I've been running 32-bit Windows XP Pro SP3 on a Intel Q6600 @2,4 GHz and compared that to my cabinet that's running 32-bit Windows XP Pro SP3 on a Pentium 4 @2.66GHz.
I started up Carnevil on both and tried diffrent builds of MameUIFX to see if some where better than others. O'course the Quad was a lot better. Running almost flawlessly.
Some skipping in the graphics but sound was playing at normal speed.

Yesterday I upgraded my Quad PC to 64-bit Windows 7 and compiled a 64-bit MameUIFX.
It's like day and night compared to running 32-bit XP.

So If you're running 32-bit OS on 64-bit CPU... UPGRADE! 
Now I have to start putting money aside for a major upgrade of my MameCab.
Cheers.
-Kman

[u_rebelscum]

Ill try to get around to doing some of your suggestions this weekend Robin.

Thanks.  Don't spend too much time on it, but I'd be interested in anything you're able to do.

Im curious as to why you would care about the non-optimized versions?...

There are 2 reasons to test the non-optimized:

- Most people (not same as BYOACers) download a precompiled version, even 64bit.  IOW, more "real world".  (I'm not sure if they get the official non-optimized from mamedev.org, or if they get an optimized one from somewhere else, though.)

- I'd like to see how much the optimized version helps.  In older (32 bit on P4 & athlons) versions, the difference was up to 5% speed increase, but 0%-2% was more normal, and a speed decrease wasn't unheard of, depending on game.  IOW, the speed increase was usually within the margin of error.  Plus the AMD optimized helped a little less than the P4 optimizations (-0.5% - 1% vs 0% - 2.5%, average IIRC). 
Is it the same years later, or have things changed?  Is the compiler better at optimization now?  Is the AMD optimization up to the same as intel optimization now?  Do some games benefit significantly more than others?  Does optimization still not make a difference for most games?
 

Anyway, comparing the nonoptimized & optimized is more of a test of the compiler than the CPUs, and how well it optimizes, and if the CPU makes a difference, and if people should bother with optimizing at all.   Not a direct test of the CPUs ability to run mame, true.

Again, not very important, but I'd like to know if what I still say in some of my posts is still true. 

I didnt know that any games used anything beyond 2 cores.  Ill definitely rerun with MT off completely, that should give a fairer comparison between the two, probably should have done that in the first place.  I imagine the use of mt should be a fairly consistant percentage no matter which processor, though I guess I can see that as well when I get it run.

I can't find the thread, but IIRC Aaron has stated a few games can benifit from 3+ cores.  IIRC it was one of the UDR (Universal Dynamic Recompiler) emulation that was able to, and only helped a handful of games (ie: <1% of the games).  So as a general rule, saying mame doesn't benefit from more than 2 cores is fine, but AFAIK there are a few exceptions.  I wish I could fine which games though.

The -ddraw flag, and in fact the entire line, was just taken straight from the 4ghz thread.

Which was used because it's great for backward comparability of older mames that only do ddraw, and what MameBenchmark uses.  But mame has defaulted to d3d since 0.107, or 3.5 years ago.

JohnIV (mameUI) benchs with -video none and -cideo d3d because the first skips any slowdown due to the video card, and the latter is mame's default which most people use. 
(BYOACers may mak up most of the ddraw'ers, though.)

I doubt there will be much difference though, so don't worry about running this if you don't want to.  (it would only be interesting if there was a diference, right? )

Vista ... XP a ... Win7 is supposed to be faster than Vista, but probably not faster than XP, speaking in incredibly generic terms, I think theres probably not much error there.

Probably right.  (only interesting if not, again )

Ill try to grab Dolphin, though its probably not in my current set since Im at 0.132...

Just wondering, since AFAIK it's the slowest game in mame ATM.


Thanks, again! 

[Deadly]

I haven't had any time to do any real benchmarking as of late but having a little time tonight I ran across this thread and thought I'd give the Dolphin compiling advantage "theory" test a go. I pretty much used JohnIV's settings ... I guess in all reality I don't feel it matters much. You just want to see a comparison between the "Simpleton Mame64 builds" VS the "BYOAC expert compiled version". <-- I had to laugh because I'm a simpleton and never tried compiling before.

So the end result: Using a stock QX6850 running the game Dolphin with a compiled MAMEUI64 ver 0.136 VS a standard release MAMEUI64 0.136 was right around 1.5% increase. Meh

IMHO for the extra work involved to compile I honestly don't see any real advantage to compiling other than skipping the nag screens. I think a person should focus on hardware optimization - I would arguably say the "need to be spot" right now is ~3.5ghz. I've run games at 3.5 and 4.2 and quite honestly I observed zero visual and audible difference in performance for instance Gauntlet legends. You drop below 3.4 and the audio starts to jah jah jah jitter   I'm not sold on that 4ghz mark but better to have too much than not enough I suppose. "I pulled these numbers on 3dmark" lol. Maybe I should bench dolphin on my 965 I7 hmmm.....   I would imagine I'd see next to zero difference as well. I will arguably guess we've hit a brick wall until the devs learn to utilize the extra cores more efficiently. Said with love of course  But then again this is coming from a person who compiled Mame for the first time tonight heh

BTW special thanks to Headkaze for using his personal time to offer the GUI tool. Very nice man.

[saurian333]

I didnt know that any games used anything beyond 2 cores.  Ill definitely rerun with MT off completely, that should give a fairer comparison between the two, probably should have done that in the first place.  I imagine the use of mt should be a fairly consistant percentage no matter which processor, though I guess I can see that as well when I get it run.

I can't find the thread, but IIRC Aaron has stated a few games can benifit from 3+ cores.  IIRC it was one of the UDR (Universal Dynamic Recompiler) emulation that was able to, and only helped a handful of games (ie: <1% of the games).  So as a general rule, saying mame doesn't benefit from more than 2 cores is fine, but AFAIK there are a few exceptions.  I wish I could fine which games though.

Correct me if I'm wrong, but when you're looking at an emulator, isn't it a question of whether the emulator itself will utilize the multiple cores?  So technically, a game that doesn't "benefit" from 3 cores vs. 2 would still be utilizing it as the emulator allows.  While there might not be a noticeable difference to the user, certain benchmarks may show a slight increase.  Obviously you don't need 3 cores for Galaga, for example, but if it's using 3 cores, there should still technically be some kind of performance increase.  Another factor to consider is how well your OS and any other background processes are utilizing those cores, and how that is impacting overall performance.

Just a thought; again, let me know if there's something I'm missing.  I don't have any systems that close to each other (e.g. a 2.8 single core to compare with my 2.8 dual-core) to run any tests myself, so I'll have to take others' word for it.

[u_rebelscum]

Correct me if I'm wrong, but when you're looking at an emulator, isn't it a question of whether the emulator itself will utilize the multiple cores?  So technically, a game that doesn't "benefit" from 3 cores vs. 2 would still be utilizing it as the emulator allows....

I think you're making the assumption that all games are emulated with the same code. 

Each emulated CPU, video chip, sound chip, etc (IOW each emulated chip) has its own code.  Sure, all games use the same mame core and OS dependent code, but little of the emulation is done in these shared code.  More important, most (if not all) of the slowdown occures in the chip emulation code.
(^This is a simplified view, but I hope not over simplified.)

So, if one or two CPU's emulation code is able to multi-thread enough to use more that 2 of a PC's cores, it will only help the games that use that CPU.  Which IIRC is what's happening, and why only a few games can effectively use more than 2 core.

[massive88]

Correct me if I'm wrong, but when you're looking at an emulator, isn't it a question of whether the emulator itself will utilize the multiple cores?  So technically, a game that doesn't "benefit" from 3 cores vs. 2 would still be utilizing it as the emulator allows.  While there might not be a noticeable difference to the user, certain benchmarks may show a slight increase.  Obviously you don't need 3 cores for Galaga, for example, but if it's using 3 cores, there should still technically be some kind of performance increase.  Another factor to consider is how well your OS and any other background processes are utilizing those cores, and how that is impacting overall performance.

Just a thought; again, let me know if there's something I'm missing.  I don't have any systems that close to each other (e.g. a 2.8 single core to compare with my 2.8 dual-core) to run any tests myself, so I'll have to take others' word for it.

To perhaps simplify and be even more incorrect, you can think of Mame as many (or multiple even!) emulators in one package.  Each arcade board basically is a collection of the individual hardware chips.  MAME emulates the chips, and the drivers put the chips together to emulate the board, in a more traditional sense (NES emulators for example have only one board to emulate, so breaking it down like this isnt all that useful).  So if you say a game benefits, you are really saying that a driver benefits, which is really saying that some of the components emulated in the driver benefit.

A chip emulated in the galaga driver may not gain anything, but a different chip emulated in the Killer Instinct Driver might.  The collections of chips make a driver, and all of these drivers are under the single MAME package.

Some games use the same driver (different software on the same hardware), but many times the hardware is unique in its application (hardware uniquely used for the software) if not components (emulated chips).  Was that adding more confusion?

[massive88]

Ok, went through some more, this time using 0.136.

The first column is the 0.136 64-bit build as downloaded from Mamedev.  Other columns are using source and Headkaze's compiler optimizing for Core2/AMD64, 2 cores, and 64-Bit.

The color in the first column is comparing the AMD to the Intel, if its red, the other chip does it faster, if its green, it was the faster chip.  In the second and third columns, they are being compared to the column to the left.  If Optimization made a sizable difference, its green.  If nomt was slower than optimization, its red.

On the AMD chip, this time I clocked it up to 221 mhz x 14.5 so its actually running at 3.2 ghz, not 3.19 it was before.  Not much of a difference, but the actual speeds of the two chips should be closer now.

I tried the -str argument Robin, but that made the first test never end.  Not sure what was going on, but they ran fine once I dropped it off.  These runs are using the same string originally done, except with D3D instead of ddraw.



Edit: Also just ran the AMD64 using ddraw instead of D3D, looks like there was no effect on performance.  Some were slower, some were quicker, all within 2% of each others numbers though, so probably within the variation of the tests themselves.

[saurian333]

I think you're making the assumption that all games are emulated with the same code. 

To perhaps simplify and be even more incorrect, you can think of Mame as many (or multiple even!) emulators in one package.  Each arcade board basically is a collection of the individual hardware chips.

You guys are right, of course.  I knew arcade ROMs were collections of multiple emulated chips, and just wasn't thinking about that when I made that statement.  Silly of me, really.

For a console emulator, I'd be correct, since the software would be emulating the entire system, and all information in a ROM is simply a data dump from a cart or CD.  Obviously the arcade ROMs are more complex.

That last set of results is very interesting.  I'm still curious as to how the OS's utilization of the cores affects things.  For example, if the OS is leaving the Phenom's third core basically untouched, and most of the games are not using it either, then it's no surprise that there's no increase.  But if you could set the affinity of most of your other processes to that third core, leaving more of the first two for MAME, I wonder if there would be more of an improvement over the Core2.  Not sure how far you can go with that; even in Linux, you can't set the affinity of everything, and probably even less so in Windows.  Even so, that may be beside the point since the Phenom X3s are generally much cheaper than the Core2s, and apparently yield at least similar performance.

[u_rebelscum]

Ausome to see the numbers!  looks like the compiler optimizations have gotten a little better than a couple years ago.  

First a clearification:

I tried the -str argument Robin, but that made the first test never end.  Not sure what was going on, but they ran fine once I dropped it off.  These runs are using the same string originally done, except with D3D instead of ddraw.

"-str" is short for "-seconds_to_run", and mame will except either form.  If you had both, there might be some issues, and if you left either blank, there will be problems.  I was trying to say, "Since you're already used -str (aka -seconds_to_run), mame doesn't show the nag screens."  I can see how my original post wasn't clear.  Sorry.


Okay, back to the numbers (see attached table).  For the averages, I get close but slightly different numbers.  (Calculated differently?)  

Anyway, the phenomII optimization averages a 3% speed increase for the tested games, with a max increase of 4.7%.  The Core2 optimization averages slightly higher, 3.7%, but has a much wider range: max increase of 11.5%, and the min is a decrease of -6.8%.  The -MT option differences are simular: Core2 aveage higher with wider range, including a decrease, while phenomII is a little smaller difference and range, but all positive.

And not to be missed, the phenomII holds it's own again the core2, Ghz for Ghz,  with four games favoring each CPU, and one game a virtual wash (< 2% difference), and the total average of the nine game at less than 0.5% difference.  All games tested except dolphin ran at 95% or greater for the phenomII, but for the core2, gradius4 was under 80%.  If these nine games are a good representation of all the games in mame, things are looking very good for phenomII:

The phenomII on average has the same bang as the core2, priced at a lower buck, for a total better bang per buck.  We also can't say "Core2 is the only way to go" anymore.

edit: uploaded second table

[Kman-Sweden]

So who'll buy a Phenom II x4 to bench? 

[saurian333]

We also can't say "Core2 is the only way to go" anymore.

Maybe my opinion (you know what they say about those), but I daresay that was never the case.  I've been running an Athlon64 X2 for 2.5 years, not overclocked.  I've compared it with friends' similar machines running Core2 of same or even higher speed.  I haven't had one significantly outperform it yet.  Higher benchmark numbers, yes.  Noticeable real-world performance, never.

The only place I'd recommend a Core2 is in a notebook.  The mobile AMD chips run significantly hotter, so they almost always have louder fans.  Kind of irritating sometimes, although a small sacrifice to get the much better value, IMO.

[massive88]

I was very sloppy with my averages, just put the average of one column divided by the average of another, which is the reason for the error between our numbers.

The games were more or less selected at random, and I think all use different drivers.  Its really interesting to me that it turned out how it did, basically as many faster on one chip than the other, which strongly points to them being more or less equal (again, given the limitations of my tests, with the AMD computer using slower ram and a different OS).

Anyway, glad to put out some numbers that might help someone make a decision down the road, and thanks for the help everyone.

Oh and also, I tried playing some gauntleg on the AMD using the 0.136.  It played fine, sound had some breaks and stutters, but graphically it was smooth as can be.  Could probably tweak it to work fine if I spent more time with it, or use 0.132 with it since it seemed to score out better.  Anyway, definitely playable.

[massive88]

So who'll buy a Phenom II x4 to bench? 

Id expect very little gain over the x3 I used, or an x2 for that matter.  As shown above, utilizing multiple cores on average gave about 3% better performance.  Unless you know of a specific game that you want to play, that specifically is coded well for multi threading, imo its not worth the price to go anything above 2 cores.

[dmarcum99]

These are very good numbers to go by.
Thank you for your time and research!  Now I'm starting to second guess my cab's performance.

Are the phenoms AMD's only "good" option for a strong mame cabinet?  NEWEGG has a good deal on a AMD 64X2 6000+ processor.  Wondering if it would be a good alternative or should I keep my conroe E4300.My mobo won't overclock the E4300....it's a factory gateway mobo (ecs 945gct-m3)

http://www.newegg.com/Product/Product.aspx?Item=N82E16819103772

[clok]

We also can't say "Core2 is the only way to go" anymore.

Maybe my opinion (you know what they say about those), but I daresay that was never the case.  I've been running an Athlon64 X2 for 2.5 years, not overclocked.  I've compared it with friends' similar machines running Core2 of same or even higher speed.  I haven't had one significantly outperform it yet.  Higher benchmark numbers, yes.  Noticeable real-world performance, never.

The only place I'd recommend a Core2 is in a notebook.  The mobile AMD chips run significantly hotter, so they almost always have louder fans.  Kind of irritating sometimes, although a small sacrifice to get the much better value, IMO.

I have two machines my old system a FX60 (oc to 2. and a E6600 (OC to 3.2) Both machines are 4 gigs of memory and 8800 cards (same card actually) and while the speeds are not  the same (.4 dif) I cant agree. I used the fx60 and it played everything wonderfully, no argument. But I had a 22" running at 1600X1050, I soon upgraded to a 24" then a 26" and soon a 30" at 2560 X1600. benchmarks show huge numbers much like you said, but I didnt notice anything when I had my 22". but as my screen increased in size and I had to bump my resup the speed problems started showing. Newer games show it more. My trusty (and obcenly expensive at the time FX chip) is holding in there even on new games on my old 22", but I do know there is a difference. But more most people I know where you are comming from, those 50% benchmarks in PCMARK matter little to the guy playing a game.

Sadly Im in the same boat as you now again. I have a 8400 OC to 4.2 and cant bring myslef to upgrade to the I7 stuff.... But as a computer nut, its going to happen. if there is anything good to say to being a technology binge buyer, i have broke one habit. the $500+ video card every 3-6 months.. wont do it anymore.

Oh and sweet benchmarks, I honestly didnt see AMD doing that well, good to see, maybe we will see the $200 sweet spot for mainline chips again since Intel has slowely ratcheted it up since they have the started getting the sales they lost to AMD back in the athlon and Athlon II days.

[saurian333]

[...] i have broke one habit. the $500+ video card every 3-6 months.. wont do it anymore.

I'm sick of upgrading my video cards at all.  I recently got a GT220, thinking it would be a great improvement over my 7900, only to find out that it wasn't.  Higher specs all-around (including twice the RAM and higher bandwidth), but some performance ratings ended up being slightly higher and others slightly lower.  I only got a slight improvement, plus DX10 capability.  I could have gotten that with a 9500 for half the price.  Seriously, a 2-year/3-generation upgrade really should make more of a difference.  I'm tired of trying to keep up; no more new video cards for me unless I'm replacing a whole system.

[massive88]

Just brought up this thread on another forum, and realized I made a mistake from square one.  The chip I was testing was an Athlon II not a Phenom II.  The only difference appears to be L3 Cache (and price because of it).  Anyway, for posterity sake, the chip I was using was a Rana, which is an Athlon II, not a Heka, which would be the Phenom II.

[Malenko]

I should run these tests tonight, I have a Phenom quad core from like a year and half/2 years ago, its a sledgehammer

[u_rebelscum]

...The chip I was testing was an Athlon II not a Phenom II....

That's actually good to hear.  Some of the benchmarks out there show little different between phenom II and athlon II, while others show ~15% difference at same speed.  So if mame is one of the case that Phenom II vs Athlon II is a big difference, then the phenom II is even better. 

But as you say, the major difference is the cache size, so maybe no big difference.

I should run these tests tonight, I have a Phenom quad core from like a year and half/2 years ago, its a sledgehammer

Thanks.  Is that a Phenom, or Phenom II?  Major differences between the two, at least on many of the other benchmarks out there.  (I wonder if it's the same with mame?)  Anyway, it would be nice to see mame athlon II vs phenom II comparison.

[Malenko]

you tell me what I got...

no time to run tests tonight, but tomorrow I will. I need to copy the roms and CHDs across the network and I havent eaten yet.

[mlalena]

Question for the OP and anyone else who has overclocked...
Most overclocks show well over 100% for gauntleg.
I am running a Core 2 Duo @ 4 GHz and I get ~80% when playing the game, and it locks up after about 2 minutes.
I've tried other similar games (like NFL Blitz) and still around 80% frame rate.
I've read that others have had the same problem but haven't seen a solution.
I'm running MAME 64 130b  (vmame64.exe) on Windows 7 64 bit.
Any ideas?

[Popcorrin]

Question for the OP and anyone else who has overclocked...
Most overclocks show well over 100% for gauntleg.
I am running a Core 2 Duo @ 4 GHz and I get ~80% when playing the game, and it locks up after about 2 minutes.
I've tried other similar games (like NFL Blitz) and still around 80% frame rate.
I've read that others have had the same problem but haven't seen a solution.
I'm running MAME 64 130b  (vmame64.exe) on Windows 7 64 bit.
Any ideas?

Alot of the benchmarks are measuring the frame rates during attract modes.  Actual gameplay creates a considerable drop in speeds.  That being said, I would still think you would get full speed on NFL Blitz with a 4GHz core 2 with mame 64, though I do not have a 4GHz core 2 to test with.   

[UberCade]

I can provide some Athlon II benchmarks once I get the system parts I ordered, which should be here by Tuesday. I'm going to overclock it as far as I can and do some testing, so if someone has a similar Phenom II system to benchmark we can compare. I chose AMD over Intel because I already have 2 Core 2 Duo systems in my house, as well as a Core i7 system and I just wanted a cheap build for my MAME cabinet. I got the whole system for $380 after tax and shipping, and I should be able to get it to 4 GHz or pretty darn close, so it should be a good comparison.

[mlalena]

OK. My bad. The mobo had a bad post and reset the overclock back to default state - 3GHz.
I think the kids repeatedly pushing the power button on the computer made it think the OC settings were bad - they like to press shiny buttons.

So now that I am back at 4GHz, the Gauntlet Legends intro / attract mode runs well over 200% and I can get 100% when playing the game, but every so often it dips below 100%. When there are a lot of enemies on the screen the game will still lock up.

I tried putting it into the AUTO frame rate skip and it took longer to lock up (got past the first level) but
the game still eventually locks up.

Are there any audio settings I can tweak or anything else that will reduce the CPU load?

[Popcorrin]

Are there any audio settings I can tweak or anything else that will reduce the CPU load?

You can reduce the samplerate but it won't gain you much.

[massive88]

OK. My bad. The mobo had a bad post and reset the overclock back to default state - 3GHz.
I think the kids repeatedly pushing the power button on the computer made it think the OC settings were bad - they like to press shiny buttons.

So now that I am back at 4GHz, the Gauntlet Legends intro / attract mode runs well over 200% and I can get 100% when playing the game, but every so often it dips below 100%. When there are a lot of enemies on the screen the game will still lock up.

I tried putting it into the AUTO frame rate skip and it took longer to lock up (got past the first level) but
the game still eventually locks up.

Are there any audio settings I can tweak or anything else that will reduce the CPU load?

Locking up may be a result of overclocking your processor too much.  Mame itself shouldnt lock up, but the processor may be taken to unstable levels, then when loaded as much as mame will load a processor, its failing.  Overclocking always makes your CPU less stable.  The goal is to keep it stable enough that you dont lock up with regularity (or hurt the hardware).

Id try different versions of Mame, even if for just that one game.  The scores posted in the OP are with Mame 0.132, and later with 0.136, between those two, it looks like 0.132 was faster for no other reason then being a different version.  Try that one and see if it does you better.  When you are close to getting a game working at full speed, just tweaking what version you use could make the difference.

Malenko - That info page doesnt make sense.  Are you overclocking?  The Sledgehammer codename belongs to the K8 family, as does the Socket 940.  The opteron Sledgehammer does not look like it should support SSE3.  The level 2 cache its reporting is consistent with the Phenom/Phenom2 line, but those should report as being SS4 capable, have L3  cache, and be either AM2 or AM3 slot.  If its overclocked, change it to defaults and see what it posts, what is says there makes little sense.  Heres where im referencing - http://en.wikipedia.org/wiki/Comparison_of_AMD_processors

[u_rebelscum]

Malenko - That info page doesnt make sense.  Are you overclocking?  The Sledgehammer codename belongs to the K8 family, as does the Socket 940.  The opteron Sledgehammer does not look like it should support SSE3.  The level 2 cache its reporting is consistent with the Phenom/Phenom2 line, but those should report as being SS4 capable, have L3  cache, and be either AM2 or AM3 slot.  If its overclocked, change it to defaults and see what it posts, what is says there makes little sense.  Heres where im referencing - http://en.wikipedia.org/wiki/Comparison_of_AMD_processors

It might be the cpu-z is too old.  1.32 came out March 27, 2006, while the phenom X4 9500 came out in November 19, 2007.

Try downloading the newest CPU-Z, 1.53 (official site, or cnet).

edit: Assuming opu-z is too old, I'd guess you have a four core phenom (one).  Another CPU I'd like benchmarked.