Vista32- vs. Vista64-bit OS Showdown *Done!*

[apoppin]

Originally posted by: lopri
I read the updated results in the first page, but I do have to question the selection of games. The merit of 64-bit OS is undoubtedly the ability to handle more memory. In that sense, the selected games are not very memory-hogging, IMO. When 64-bit OS show its benefit, if it does, will be when a game attempts to access huge amount of memory. (Company of Heroes coming to my mind) Half-Life 2 or F.E.A.R. aren't the best title to show the benefit of 64-bit OS.

It's kinda different issue but the most memory hogging games tend to be RTS and RPG games, not FPS games. And it's kinda natural if you think about it. The size of world and number of units/items grow as the game progresses, and that's when the extra memory comes in handy. I have played Company of Heroes both on 32-bit Vista and 64-bit Vista, and on a large map where a hundreds of units are duking out, the advantage of 64-bit and extra RAM (I had 8GB at that time) was apparent. Unless a game ends in its early stage, 32-bit Vista invariably crashed later when there are hundreds of units duking out on a gigantic map. Did not occur on 64-bit Vista.

that is why i am hoping someone else will join in ... the closest thing to RTS i played and enjoyed was B&W2

i have picked the popular and relatively current games that i own. i'd love to see someone bench CoH.

i have the Witcher and the most memory eating game of all - Gothic3 .. i don't believe there are any benches already pre-made on the Internet as they are for all the other games i am benching. If you can find anything, LMK

and i have completed Vista32 benching of STALKER and PREY ... Stalker seems a bit smoother than i remember in May - and Prey runs worse


i am gonna reboot and run the 64-bit benches and post them

back ,, in a few ...

-when it's done

 

[imported_wired247]

sweet... I just got into the witcher starting last night. very interested to see the results

 

[apoppin]

Ok, a quickie update before i either set up CrossFire or drop from exhaustion
[everything @16x10]

PREY ... *everything maxed* and "boost" plus 4xAA/16xAF

64 bit Vista
Min Max Avg
41 120 77.852

32 bit Vista
Min Max Ave
44 124 80.446


==================


....and STALKER *totally* maxed* in game settings and Full dynamic lighting
{all sliders to the right}


Short Demo

64 bit Vista
Memory > Min Max Avg
262118 K > 15.03 556.01 63.60

32 bit Vista
Memory > Min Max Avg
259620 K > 18.52 636.08 68.88


Buildings Demo

64 bit Vista
Memory > Min Max Avg
233492 K > 10.83 712.31 70.84

32 bit Vista
Memory > Min Max Avg
226107 K > 11.73 722.00 71.57

i am betting on exhaustion ... i'll update the 2nd post tomorr ... later

g'nite
:moon:

 

[apoppin]

i thought i'd just toss in that 2nd card

why can't everything be easy? ... where is that damn Staples button?
---i have to move a LOT of crap - everything - to SQUEEZE that card into the 2nd PCIe slot
--'cooling' looks *fun* ... i am beginning to be glad it is NOT a 2nd 2900xt

and i need 2 Crossfire bridge interconnects!?! ... it looks like i need two, right?
- i have only *one* from my 2900xt box -G*D* Sapphire didn't bother to include it ... :|

or am i wrong?


http://www.sapphiretech.com/us..._overview.php?gpid=200

Dual channel interconnect is not required for ATI CrossFire, and may not be included in all product configurations

 

[apoppin]

So far ... 9/10 Vista32 wins in Part 1 of the FPS testing

*except FarCry* - It's a blowout ... Vista64 wins by a huge amount
- do you think FC is optimized for 64-bit ... if all games played like FC i'd dump Vista 32 in the garbage

i'm on to the last one ... Lost Planet ... i need to get the 32 benches and run the 64-bit again ... then it's Xfire and a couple of comparisons

here are two more:

Painkiller v1.5

c1l1 Benchmark

64 bit Vista
Minimum Average Maximum
31.27, 204.06, 333.37

32 bit Vista
Minimum Average Maximum
33.31, 210.74, 334.37


c1l2 Benchmark

64 bit Vista
Min. Average Maximum
7.24 185.72 512.00

32 bit Vista
Min. Average Maximum
6.95 189.88 512.00



So far there is *nothing* that recommends Vista64 over Vista 32

*EXCEPT:

FarCry - 64-bit

Resolution: 1680 × 1050 (Custom)
Ultra quality option, Direct3D renderer
Level: Pier, demo: 1.tmd
Pixel shader: default model
Antialising: 4×
Anisotropic filtering: 16×
HDR: disabled
Geometry Instancing: enabled
Normal-maps compression: enabled

Score = 101.14 FPS (Run 1)
Score = 100.37 FPS (Run 2)
Score = 100.88 FPS (Run 3)
Score = 101.37 FPS (Run 4)
Average score = 100.93 FPS -64bit

Frames, Time (ms), Min, Max, Avg
3471, 34361, 67, 152, 101.016 -64bit


FarCry - 32-bit

Frames, Time (ms), Min, Max, Avg
3501, 58323, 42, 92, 60.028 -32-bit

Resolution: 1680 × 1050 (Custom)
Ultra quality option, Direct3D renderer
Level: Pier, demo: 1.tmd
Pixel shader: default model
Antialising: 4×
Anisotropic filtering: 16×
HDR: disabled
Geometry Instancing: enabled
Normal-maps compression: enabled

Score = 60.02 FPS (Run 1)
Score = 59.34 FPS (Run 2)
Score = 59.59 FPS (Run 3)
Score = 59.08 FPS (Run 4)
Average score = 59.50 FPS 32-bit

IF this is right ... and32-bit Vista results are in-line with my testing of FC back in May with settings at 16x12/14x9 - the we see the *future* ... games optimized for 64-bit will be much faster than 32-bit/32-bit OS

Look at the benches again side by side:

32-bit Average score = 59.50 FPS
64-bit Average score = 100.93 FPS

Frames, Time (ms), Min, Max, Avg via FRAPS
3501, 58323, 42, 92, 60.028 -32-bit
3471, 34361, 67, 152, 101.016 - 64bit

 

[apoppin]

*Done* ... at least with Part One - FPS comparison - 9/10 benches belong to Vista 32 - 1 bench, FarCry belongs to Vista 64 ... and that one is a blowout! If they were all like FC, Vista32 would be in the trash.

i just set up CrossFire with my 2900xt and brand-new 2900pro ... Damn i had to move *everything* around to keep some airflow [and my HDs were in the way]; WORST of all, if i keep it, i need to reconfigure my ... well, start over again ...
- first of all, the 2nd card covers two of my SATA ports and i will now have to go back and use the "purple" RAID ones
:roll:

OK, it is SO easy setting it up [assuming you have room and don't have to tear down and rebuild like i did this evening] ... stick in the 2nd card, plug in the 2 bridge interconnects [i have only one; Cheap Sapphire doesn't include it and i had to order another from NewEgg, arriving Tuesday] ... and boot up. Windows recognizes crossfire and you have to reboot. That's it .. you're done.

Of course, i only have one interconnect so my bandwidth IS limited ... but CrossFiring with the Pro brings my 3DMark06 score up from 10094 with Vista 64 to 12251 ... and that is the first run ... i am also hoping for *better* when i add the 2nd Bridge Interconnect and OC my CPU back to over 3.15 Ghz ... looks like maybe 25% increase ... and if i OC it, it will blow an Ultra out of the water
... i feel like bush ... "mission accomplished"
:Q

:laugh:

at least it isn't any louder
:shocked:

here the Lost Planet results ... damn near *identical* in every way ... best 2 of 4 runs each:


Lost Planet: Extreme Conditions full retail game built-in demo. DX10/everything fully maxed in-game/1680x1050/4xAA-16xAF

Snow - Vista 64 - 19.6/19.5 / Cave 28.1/27.0
Snow - Vista 32 - 19.6/19.4 / Cave 28.0/27.2

both bottomed just over 11.1 FPS and topped at 30.2 FPS in Snow, the more demanding of the two benches.

i'll see you in the AM .. i am gonna play with Xfire before i pass out
:moon:

 

[dileepsv]

why do u guys rely on 3dmark when almost everyone here says its useless and that its shit?

 

[apoppin]

Originally posted by: dileepsv
why do u guys rely on 3dmark when almost everyone here says its useless and that its shit?

who is *everyone* ?



check *almost every* major review site ... *they* use it
-it has it's 'uses'

and why are you fixated on one out of ten benchmarks?

 

[UTFan81]

Originally posted by: UTFan81
Actually I've been here a while I just made a new account for privacy reasons.

No can do. This account is locked. It's against the rules.

esquared
Anandtech Senior Moderator

I'm back. so when are you doing part 2 and what will it be based on?

 

[ViRGE]

Originally posted by: apoppin

Originally posted by: dileepsv
why do u guys rely on 3dmark when almost everyone here says its useless and that its shit?

who is *everyone* ?



check *almost every* major review site ... *they* use it
-it has it's 'uses'

and why are you fixated on one out of ten benchmarks?

Apoppin is right. 3DMark is a terrible choice for when you want to do cross-vendor comparisons because of all of the damn little micro-optimizations everyone does to get that extra point, but if you're comparing the same vendor (or better yet the same exact hardware) then it's a perfectly viable benchmark.

 

[apoppin]

Originally posted by: UTFan81

Originally posted by: UTFan81
Actually I've been here a while I just made a new account for privacy reasons.

No can do. This account is locked. It's against the rules.

esquared
Anandtech Senior Moderator

I'm back. so when are you doing part 2 and what will it be based on?

so you straightened it out ... good

Part two? You think we *need* part Two?


everything so far has supported my original PoV .. that games are written for 32-bit with the 2GB barrier well in mind. The games that "eat" RAM were just badly written as they goofed up with Vista being so new - CoH, the Witcher, etc. - as an example Hg:L had serious problems with RAM management - they fixed it and DX10 now looks and runs better than DX9 with the same settings. *Generally* Vista 32 leads Vista 64 in FPS - except for FC [which i believe is "64-bit friendly" - one out of 10,000 games]


:Q


Just Kidding ...
[sorry, man ... 5 hours of sleep a night doesn't really cut it for me - and i am back to work tomorrow. You know, i really didn't intend for this to turn into a SOLO project. ]

This week ... i need a little break, and i am *hoping* nullpointerus has his rig back to perfection [finally] this morning ...

i have seen so many of our "elitists" talk about how Vista 64 "feels" faster - BullSh!t - from my *limited experience* with both OSes side-by-side with 32-bit games - there is nothing to recommend Vista 64 over 32-bit.
... and Of course, there is absolutely nothing wrong with Vista64 [whatsoever]

toss a coin ... or pick Vista 64 for "future-proofing" ... it doesn't really matter as Vista Premium is only ~$110 - or buy Ultimate for a few bucks more ... it isn't like making a choice for a future wife; i'll be glad to 'toss out' my 32-bit copy in a couple of years for 'ultimate' - or for an entirely hot 'new' wife.

------------------

Originally posted by: ViRGE

Apoppin is right. 3DMark is a terrible choice for when you want to do cross-vendor comparisons because of all of the damn little micro-optimizations everyone does to get that extra point, but if you're comparing the same vendor (or better yet the same exact hardware) then it's a perfectly viable benchmark.[/quote][/quote]


--Thanks, and Especially in this case ... where *everything* is 'identical' [as humanly possible]. 3DMark06 tracks "changes" in my system ... and in this case is a pretty good measure of the performance increase by adding a 2nd GPU
--up from 10094 with Vista 64 to 12251 ... which is about 'right' when comparing performance in actual "games" ... a definite noticeable improvement in the games i played [before collapsing from exhaustion early this AM]


addendum

..... Btw, Crysis Demo in CrossFire looks like *crud* in Vista 64 - texture crawling and shimmering - it "looks" like sloppy AFR with a ATI MAXX.
-Vista 32, otoh, is *perfect* with Crysis Demo ... Still Slow - i won't be playing DX10 16x10 and "very high" .... but it is as perfectly rendered as with a single GPU.

take you pick ... one is not more high-end than the other for gaming ...
:music: ... and the beat goes on ... :music:




expect Part Deux ... later
--when it's *done* .

... and someone else, please ... it isn't at all hard [time consuming, yes] ... you will have a "brand-new" squeaky-clean rig and the latest benchmarks ... and you won't have to listen to anyone give you FUD about your choice of OS.



and i am very OPEN as to the *how* we can do Part 2 ... i have my own ideas ...

 

[UTFan81]

Vista 64 does feel faster. That doesn't mean it's faster for gaming though.

 

[SniperDaws]

Who cares both Vista 32bit and 64bit arent worth installing, and by the look if Windows 7 that wont be worth playing with either.

 

[apoppin]

Originally posted by: UTFan81
Vista 64 does feel faster. That doesn't mean it's faster for gaming though.

Yes ... agreed... Vista 64 IS faster for some things [period]. Remember, we are testing 4GB gaming rigs. And it looks like i am at the halfway point, or so.

Look at FarCry ... i believe it IS optimized for 64-bit and the performance increase over Vista32 is undeniable. i can certainly tell the difference in the "smoothness" even though Vista 32 ran it well above 30FPS - if even 3 out of the dozen games i experienced were like that, Vista32 would be SO uninstalled and i would dual-boot with XP32/Vista64.

As it is ... unless Part Two of the testing gives me a big surprise, it looks like Vista 64 gets the boot.


:Q

i mean Vista 64 gets booted OFF my My rig ... i doubt i will *ever* use it as the next OS from MS will probably be my choice for cutting edge gaming in 2011 or so. For now and probably for the next 2-3 years, Vista 32 and XP will share the 'boot' on my rig. So far, i am pretty confident that Vista 32 is still very much "high end" for gaming.

And ... again ... P-L-E-A-S-E ... i am taking a little break from formal testing to informally test my new CrossFire toy ...
--in the meantime ... please feel to contribute suggestions for Part Two - the 'how" you would like to see it tested.

Mahalo!

both Vista 32bit and 64bit arent worth installing

XP is SO primitive in comparison ... and Vista3-2 runs DX9 games just as well as XP
... otoh XP doesn't run *any* DX10 pathways - don't be fooled by the "Crysis looks as good on XP" nonsense - it doesn't[period]
... and i AM totally enjoying DX10 - especially Hg:L - now that they fixed DX10 to look AND run better than DX9 with the same [fast] HW. i expect other games will follow suit.

 

[JustaGeek]

Great job, apoppin!

Or just a confirmation of what we have known all along...

BTW, which version/patch of Far Cry did you use...?

I am now using 1.3, since all the later ones were specifically "32-bit", and I tried the 64-bit AMD patch without success (the game would just not run with the 64-bit patch...)

Did you actually patch it to 1.4...?

TIA

:thumbsup:

 

[apoppin]

1.4 is the minimum that would work with my FC benchmarking tool ... i have no knowledge if it is 64-bit ... all i know is that FC ran a hellofalot better than on the Vista32 partition.


... and thanks ... i think


now it's time to *play* some games ... and see if i got my $150's worth out of Xfire

 

[CP5670]

It's strange that Far Cry, a game that is now four years old, shows such a big improvement though while recent games show hardly any difference at all. The 64-bit version of FC was actually made for XP 64 if I remember right.

Vista 64 would be out for me personally, due to that absurd limitation with the signed drivers.

 

[apoppin]

Originally posted by: CP5670
It's strange that Far Cry, a game that is now four years old, shows such a big improvement though while recent games show hardly any difference at all. The 64-bit version of FC was actually made for XP 64 if I remember right.

Vista 64 would be out for me personally, due to that absurd limitation with the signed drivers.

i haven't checked to see if FC is ported to 64-bit for sure ... it appears it is.

i had only one problem with unsigned drivers - my bluetooth adapter. Which pretty much rules out Vista 64 for me if i want faster dial-up and no workaround. Of course there ARE workarounds .. Even 'Easy BCD' has options to disable driver signing requirements in Vista-64

 

[nullpointerus]

Just a quick update:

I've been quiet for a while now, but I've done the following with my rig:

-- reinstalled primary and test partitions
-- ordered an 8800GTS w/ Crysis, due to arrive Tues.
-- finally reached a stable o/c of 3.0 GHz ... *crosses fingers*
-- ordered 5x140mm fans for my CM690 and a GeminII CPU cooler (2x120mm)

:Q

I've been running on an X1550 P) for a while now, since my HD2900XT causes issues in my PC. I actually wanted to try gaming on this card (just for kicks) but never did find the time. My life (and my room) have been a wreck lately due to all the PC troubleshooting and a variety of things going on (e.g. cleaning/painting the garage), plus normal life.


Look for some preliminary nVidia-based benchmarks and my rig specs on Tuesday. :thumbsup:


I'll finally be getting back in this thing!

 

[apoppin]

Originally posted by: nullpointerus
Just a quick update:

I've been quiet for a while now, but I've done the following with my rig:

-- reinstalled primary and test partitions
-- ordered an 8800GTS w/ Crysis, due to arrive Tues.
-- finally reached a stable o/c of 3.0 GHz ... *crosses fingers*
-- ordered 5x140mm fans for my CM690 and a GeminII CPU cooler (2x120mm)

:Q

I've been running on an X1550 P) for a while now, since my HD2900XT causes issues in my PC. I actually wanted to try gaming on this card (just for kicks) but never did find the time. My life (and my room) have been a wreck lately due to all the PC troubleshooting and a variety of things going on (e.g. cleaning/painting the garage), plus normal life.


Look for some preliminary nVidia-based benchmarks and my rig specs on Tuesday. :thumbsup:


I'll finally be getting back in this thing!

You made my day ...

... i was feeling pretty all *alone* here


Good luck on the stable OC ... i won't have my 2nd Interconnect Bridge till Thursday, so i will resume testing "part 2" this weekend. And we can now see what a nvidia GPU adds to the mix.

now i got some *games* to catch up on



after work
:Q

 

[ChronoReverse]

Haha, good information bad presentation. Gotta work on formatting a bit to make it easier to read.

Anyways, this just confirms what I've experienced: gaming under Vista 64 is a bit slower than Vista 32.

 

[CP5670]

Originally posted by: apoppin

Originally posted by: CP5670
It's strange that Far Cry, a game that is now four years old, shows such a big improvement though while recent games show hardly any difference at all. The 64-bit version of FC was actually made for XP 64 if I remember right.

Vista 64 would be out for me personally, due to that absurd limitation with the signed drivers.

i haven't checked to see if FC is ported to 64-bit for sure ... it appears it is.

i had only one problem with unsigned drivers - my bluetooth adapter. Which pretty much rules out Vista 64 for me if i want faster dial-up and no workaround. Of course there ARE workarounds .. Even 'Easy BCD' has options to disable driver signing requirements in Vista-64

Someone here said that you can bypass it if you hold down some key while it's booting, but you have to do that every time. That would still be pretty annoying.

I have to use an unsigned monitor driver to force refresh rates effectively.

 

[apoppin]

Originally posted by: ChronoReverse
Haha, good information bad presentation. Gotta work on formatting a bit to make it easier to read.

this isn't the final ... expect a "neat package" after the Weekend

 

[DaveBC]

I'm certainly enjoying your work. :beer:

You are doing what I don't have to skills or understanding to do. I hang on every test. Looking forward to your RAM/loading anaylsis.

 

[apoppin]

Originally posted by: DaveBC
I'm certainly enjoying your work. :beer:

You are doing what I don't have to skills or understanding to do. I hang on every test. Looking forward to your RAM/loading anaylsis.

well thank-you ...



-- and flattery will get you everywhere


Seriously, it appears that a lot of computer enthusiasts think it takes some special understanding or special skills .... actually it DOES take *one* quality we all possess, considering most of us build our own rigs and toubleshoot them: P-A-T-I-E-N-C-E

We all did "experiments" in High School ... this is also very little different then cooking from a recipe ... i started long ago measuring and testing audio equipment - this is a LOT easier.

If *anyone* is still interested in joining us, shoot me a PM ... i'll help explain the "mysteries" of benchmarking ... and you can *always* do it yourself - from then on. And you will see for yourself what professional reviewers see and get a good "feel" for what they are talking about. Perhaps best of all - you may be able to judge and separate the BSers from the good reviewers.

also ... again ... i am looking for *ideas* to test Part 2
:light:

you may not like the way i am gonna do it