One thing that can't be denied about Crytek's demo of Crysis is that it's very newsworthy! A sharp eyed forum regular (cheers UTGeneral!) noticed that Legion Hardware had posted a brief comparison between how well Crysis runs in 32 bit and 64 bit modes. Their conclusion was that 64 bit was quite a bit better. Well that piqued my curiosity, so we gave it a bash too! Hmm, the results are... different...

To begin with, we took the following system and ran the CPU and GPU benchmarks in 32 bit and 64 bit mode:

• Windows Vista Ultimate x64
• Intel Core 2 Quad Q6600 @ 3GHz
• 4GB system RAM
• NVIDIA GeForce 8800 GTX 768MB
• ForceWare 163.69 drivers

We took two extremes to see what the results would churn up: 800 x 600, running in DX9 with all detail settings at Low, and 1680 x 1050, running DX10 with all detail settings on Very High. Each benchmark batch file was run 5 times, with all of the collated results averaged. The results are in the image to the right.

With our testing, on maximum DX10 settings, when the tests are essentially totally limited by the graphics card and CPU power, there's no real noticeable difference between 32 and 64 bit - ignore the fact that the minimum values do show some differences as these figures were highly variable; any background operation would just mess things up. Take a look at the low resolution and settings figures though: on such a system, the tests were a cakewalk and this time the numbers were consistent and always had the 64 bit mode running slower than 32 bit. Now this is, of course, just one system and one operating system, so we gave it another bash, using the following machine:

• Windows XP Professional x64
• Intel Core 2 Duo E6400 @ 2.1 GHz
• 2GB system RAM
• ATI Radeon HD 2600 Pro 512MB
• Catalyst 7.10 drivers

Now it would have been better to take the exact setup of the first machine and just use a different operating system, but hey - time's money and there are masses of reviews to do! Perhaps using a weaker setup would be preferable anyway: if the 64 bit version did run considerably better than the 32 bit one, there surely it would help out here?

Take a look at the figures to the right. Again, ignore the minimum values; the tests ran sooooo slowly on the maximum settings, that there's just no difference between 1 and 0.8 fps - that's just dead slow! But look at the comparable settings to first system: DX9 Low at 800 x 600. Here we see that the 64 bit mode is still slower, albeit by a tiny fraction, than the 32 mode.... nearly. The High GPU test persistently gave the same sized gap with each run though, so there we have it: 64 bit faster than 32 bit, but only as a minimum value - the average and peak were essentially the same.

So what's going on here? Didn't Crytek say that "64-bit in general runs better than 32-bit. In fact I would recommend gamers run 64-bit only under very high configurations"? Okay, we've only examined two different systems but because they are so different, surely one would expect the 64 bit mode to show some improvement, beyond the isolated case of the DX9 High GPU test? Some people on our forums have reported that the demo "feels smoother" when playing in 64 bit but we couldn't verify that directly with some in-game testing.

Before we all jump to any conclusions though, bare in mind that this is a demo and not the final retail game - we could genuinely see noticeable gains by using the 64 bit version but if we assume that the demo is a pretty close code version to the launch one, then not all of us are going to be so lucky. Please let us know what you find out: try running the CPU and GPU benchmarks across a range of settings in 32 and 64 bit, and then post your findings!

Then again, wasn't the Very High setting only supposed to be possible in DX10? Hmm....

Comments

	Dibrom 2007-10-30 #1 Disappointing Also the yougamers watermark covers the screen res. The only thing I could think of is Nick's using older drivers (ForceWare 163.69 vs 169.01). Still shouldn't matter unless the previous drivers were problematic. But then they wouldn't have about the same performance...
	Neeyik 2007-10-30 #2 The resolution is in the text: the "Low" is 800 x 600 in both cases, whereas the "High" is 1650 x 1080 for the Vista machine, and 1600 x 1200 for the XP one. If I have chance tomorrow, I'll give it another look with the new beta drivers that NVIDIA have just released.
	Pirelly 2007-10-30 #3 Nah, it isn't the drivers. 169.04 doesn't help either. for me, the 64bit gpu bench is slower than 32 bit. It's about a 4-5fps difference (28 vs 33fps). However, the cpu bench shows about the same results...(31fps average). This is at all High, post processing and water at Very High and 1680x1050 res.
	Marvin_The_Martian 2007-10-30 #4 Ok. since a> cpu benches don't change with x64 ( where the biggest gains should be seen imho ) and b> I can't see this demo using all my cores and I'm seeing this posted all the time, I'm calling this beta very rushed! Think their engine really isn't ready, or they left out parts which determine how many threads can be run. When cpu utilization is fixed, I think the x64 will also shine. As in the higher lows, that's a nice result already and I think caused because x64 allows for faster calculations in the physics thread mostly. Edit: Did anyone with a quad already checked how many threads are spawned? I didn't bother to check if it uses more then one thread but I can see my cpu utilisation on one core being around the 20/30% mark and all other just iddle and peak to 15% once in a while. That doesn't seem right to me...
	Unregistered 2007-10-30 #5 I've noticed this as well on my system: Vista x64 Ultimate AMD Athlon X2 4400+ Asus A8N32-SLI Deluxe 2GB OCZ Platinum DDR500 Nvidia 8800 GTX Sound Blaster X-Fi Platinum Regardless of settings I use 64bit crysis comes in 1-3fps slower on minimum, average and max for both the GPU and CPU benchmarks. I was honestly expecting Crysis to murder my CPU but I pull a constant 40-50% usage total between the two cores so it doesn't seem to be utilizing dual cores properly. I'd much rather see 100% usage and have higher fps and not be able to run as much in the background.
	Pirelly 2007-10-30 #6 Quote: Originally Posted by Unregistered I've noticed this as well on my system: Vista x64 Ultimate AMD Athlon X2 4400+ Asus A8N32-SLI Deluxe 2GB OCZ Platinum DDR500 Nvidia 8800 GTX Sound Blaster X-Fi Platinum Regardless of settings I use 64bit crysis comes in 1-3fps slower on minimum, average and max for both the GPU and CPU benchmarks. I was honestly expecting Crysis to murder my CPU but I pull a constant 40-50% usage total between the two cores so it doesn't seem to be utilizing dual cores properly. I'd much rather see 100% usage and have higher fps and not be able to run as much in the background. Yeah, same here... But when I fire up the UT3 demo, both my cores are at a constant 100% Shouldn't it be the same for crysis???
	ajbarnes 2007-10-31 #7 I have XP64 / Vista64 on the same system, so I may install the demo on the XP drive to see if there is a performance difference. The only striking variable is the hard drive - my Vista drive is a 16MB cache Raptor (150GB), while the XP drive is an 8MB cache Raptor (74GB). Both OSes have minimal startup programs and unnecessary background processes are disable, so this may be a good test. Of course, I may have to download the demo install file again...
	Marvin_The_Martian 2007-10-31 #8 Quote: Originally Posted by ajbarnes I have XP64 / Vista64 on the same system, so I may install the demo on the XP drive to see if there is a performance difference. The only striking variable is the hard drive - my Vista drive is a 16MB cache Raptor (150GB), while the XP drive is an 8MB cache Raptor (74GB). Both OSes have minimal startup programs and unnecessary background processes are disable, so this may be a good test. Of course, I may have to download the demo install file again... That would be a nice test for sure, I been wondering about xp64 performance with the dx10 hack
	ajbarnes 2007-10-31 #9 I'll emulate Nick's tests and add a couple of 1920x1200 tests as well, and I'll give the "very high settings on XP" hack a try too, at the same resolutions as my Vista tests. Results sometime tomorrow.
	Unregistered 2007-10-31 #10 this is funny. guys at legion hardware made the same comparison, check out the results: http://www.legionhardware.com/document.php?id=698&p=7
	Dirk_Diggler 2007-10-31 #11 Quote: Originally Posted by Unregistered this is funny. guys at legion hardware made the same comparison, check out the results: http://www.legionhardware.com/document.php?id=698&p=7 High quality in DX10............. So running DX9 then
	Neeyik 2007-10-31 #12 DX10 can be used at all quality levels and to the unregistered chap/chapess, read our news item carefully... especially the opening paragraph!
	NoDKane 2007-10-31 #13 Oh who to believe?!?
	ajbarnes 2007-10-31 #14 Here are my numbers. I don't have time to Excelify these, so here's the raw data. Interpret and twist it as you see fit. (If I get time later today, then I'll install on the WinXP partition and give that a go with the "very high" settings config edit). I let each run of the benchmark complete at least two runs before exiting out and recording the numbers from the third run. The GPU numbers aren't very interesting, though the CPU numbers show a measurable difference in 32 vs. 64 bit with regards to minimum and maximum framerates. The 32-bit CPU numbers show a higher minimum framerate and lower maximum framerate, where the 64-bit CPU numbers show a lower minimum framerate but higher maximum framerate. Averages are similar across the board, however. System: Core2Quad @ 3.55GHz 8800GTX @ 675MHz Core / 1050MHz Memory 2GB RAM @ 1150 MHz Windows Vista Ultimate x64 NVIDIA 169.04 Beta Drivers Driver settings: Single Display; No AA or AF; LOD Clamped; Texture Filtering: Quality (Framerate: MIN / MAX / AVERAGE) 32-Bit - CPU 1680x1050 - 10.75 / 24.73 / 16.56 1920x1200 - 9.30 / 21.58 / 13.9 32-Bit - GPU 1680x1050 - 14.51 / 22.98 / 18.53 1920x1200 - 11.35 / 17.97 / 14.76 64-Bit - CPU 1680x1050 - 8.06 / 26.54 / 16.82 1920x1200 - 7.40 / 20.83 / 13.06 64-Bit - GPU 1680x1050 - 14.74 / 22.61 / 18.89 1920x1200 - 10.67 / 16.71 / 14.69
	Neeyik 2007-10-31 #15 So, just taking those that are comparable to those that I did: CPU - 1680 x 1050 32 bit - 10.75 / 24.73 / 16.56 65 bit - 8.06 / 26.54 / 16.82 GPU - 1680 x 1050 32 bit - 14.51 / 22.98 / 18.53 64 bit - 14.74 / 22.61 / 18.89 To all intents and purposes, they're the same figures, apart from the CPU test minimum; not 100% conclusive evidence, but this is now 3 different machines that are showing no really improvement or "better gaming" in 64 bit.