Hardware specs?

[Steiner14]

I'm wondering if the graphic-cards specs have already been nailed down?

Will highend cards be necessary for absolute fluid high frame rates at highest details @large maps?

How does CM:BN performance-wise compare to CM:A, CM:SF?

How much power will good looking water suck up?

Nvidia or ATI?

I'd prefer to have the computer prepared and upgraded, when the game will be released.

[scottie]

I was worried about this to after having to find a replacement when my 8800GT failed. I posted the following thread on CMSF tech support and was delighted to get a reply almost instantly.

http://www.battlefront.com/community/showthread.php?t=94556

"None that I specifically know of at this point. The AMD/ATI Catalyst drivers seem to be fine at this point with CMSF, so CMBN should be in the same situation too.

"he architecture for the 69xx series Radeons is a bit different than the ones that came before it. This potentially could lead to differences in the behavior of the card with some drivers compared to other, earlier video cards using the same driver. This can be true of nVidia video cards too.

So right now no issues seem to be known with the 10.12 Catalysts using earlier video cards."

69xx use the new "Cayman" chips i subsequently found out , hopefully should be ok.

[Steiner14]

scottie, thank you.

But what card-class is necessary for maximum details and fluid frame numbers @big battles?

[Schrullenhaft]

No hardware specs have been listed for CMBN as of yet. However it is probably somewhat safe to assume that whatever works for CMSF will typically work for CMBN.

CMBN is only a 32-bit application, so it will not really use any more than 2GB of RAM itself. So a 4GB RAM system would be enough if you were only concerned about CMBN. A 64-bit Windows OS will NOT be necessary either, but it should work fine with CMBN. You'll possibly want a 64-bit version of Windows if you have more than 3 - 4GB of RAM in your system.

A fast CPU is always handy and the CMx2 can almost always utilize a faster CPU. However CMBN will NOT make use of more than one core on a multi-core CPU. So if you were considering the purchase of a new CPU, then a VERY BROAD recommendation would be to get one with a higher clock speed rather than more cores. However there are more differences between CPUs than just clock speed and the number of cores. There aren't any specific advantages between AMD and Intel regarding CMBN, though some of the Intel CPUs can be faster, but much more expensive than AMD CPUs.

As for video cards, this Graphics Card Hierarchy Chart helps to compare the AMD/ATI Radeons with Nvidia GeForces (and Intel integrated video). Even if the video cards are on the same tier, they're NOT exactly equivalent to each other, but they're roughly close for the purposes of this chart. Unfortunately there are no details on the performance differences between each tier, so this chart is mostly useful for comparing equivalent video cards between the major manufacturers. A lot of these video cards are no longer being sold new, but the chart does give you a general idea where your video card may stack up compared to others. For CMBN I would suggest at least being at the level of the following tier or higher:

GeForce 6800 Ultra, 7600GT, 7800GS, 8600GS, 8600GT (DDR3), 9500GT (DDR2)

or

Radeon X800 XT/PE, X850XT/PE, X1650XT, X1800GTO, HD 2600XT, HD 3650 (DDR3), HD 3670

CMBN WILL RUN WITH SLOWER VIDEO CARDS, but there will possibly be sacrifices in terms of model and texture quality, etc. Nothing is quite set in stone and users have different ideas of what is and isn't acceptable in terms of video performance. Large battles with larger maps and larger numbers of units might not be fully 'fluid' to all expectations, even with a fast system. CMBN is still in BETA and thus it is not fully finished to help determine what type of performance you will get with particular video cards. Again, CMSF should be a somewhat accurate gauge as to how well CMBN will play, with the exception that there will be more and larger trees, more detailed buildings and possibly more textures overall with the game.

Another caveat of the above chart is that it may not be exactly descriptive of the differences in performance when it comes to CMBN. The water effects are not going to be of the same quality and features as a number of FPS games, but they'll be above what was in the CMx1 series.

[Steiner14]

Schrullenhaft,

thank you very much.

Would you recommend ATI or NVIDIA? (i want beautiful FOG)

Since i don't want to spend much money, would you rate the speed of a Radeon HD5770 fast enough for big battles?

[Redwolf]

The information that more RAM wouldn't utilized just because it's a 32 bit program is still incorrect. You mix up virtual and physical memory and in any case the barrier is at 3 GB for a 32 bit program in a 32 bit kernel and just short of 4 GB when running in a 64 bit kernel.

Now, if the game doesn't have more than 2-3 GB data to shuffle around obviously you don't need more. But that's a different thing.

[Thomm]

Now, if the game doesn't have more than 2-3 GB data to shuffle around obviously you don't need more. But that's a different thing.

I saw little to no influence on performance in CM:SF when I upgraded my old computer from 2 GB to 4 GB.

As for CM:BN: I have a mid range laptop (video card GForce GTX 285M) and it appears to work pretty well with the beta code (I did not see any reason not to use "Best" settings so far, even on "Huge" maps).

What always impresses me is that even large forests do not seem to have an impact on frame rate (FPS difference 1 to 3 when tree display is toggled) while looking great.

Shadow rendering appears to be the most expensive item.

Apart from that, there are many factors to consider, including the amount of action going on.

Best regards,

Thomm

[Steiner14]

Thomm,

your CPU?

[Apache]

I do sometimes wonder if I'm going OTT putting a GTX 570 in my new build now given that I could put in an GTX 250 or a GTS 450 (still way above the suggested 7600 etc) for a £100 or so less than a GTX 570, still have a spec that will eat CM (with a 1GB memory) given that the faster newer cards I think tend to offer minimal performance hike over their predecessors at maximum price hike. In 3 years time or so I can then buy another GPU at around the £100 mark which will likely be some replacement for the 570 launched in two years time.

[Thomm]

Thomm,

your CPU?

Sorry that I am so late with my reply:

Intel Core i5 M 460 @ 2.53 GHz

4 GB RAM

Windows 7 Home Premium 64 bit

Best regards,

Thomm

[Apache]

It sure seems to get to a point of cut off where further expense, CM wise, may even actually be wasted.

[Knaust1]

No hardware specs have been listed for CMBN as of yet. However it is probably somewhat safe to assume that whatever works for CMSF will typically work for CMBN.

CMBN is only a 32-bit application, so it will not really use any more than 2GB of RAM itself. So a 4GB RAM system would be enough if you were only concerned about CMBN. A 64-bit Windows OS will NOT be necessary either, but it should work fine with CMBN. You'll possibly want a 64-bit version of Windows if you have more than 3 - 4GB of RAM in your system.

A fast CPU is always handy and the CMx2 can almost always utilize a faster CPU. However CMBN will NOT make use of more than one core on a multi-core CPU. So if you were considering the purchase of a new CPU, then a VERY BROAD recommendation would be to get one with a higher clock speed rather than more cores. However there are more differences between CPUs than just clock speed and the number of cores. There aren't any specific advantages between AMD and Intel regarding CMBN, though some of the Intel CPUs can be faster, but much more expensive than AMD CPUs.

As for video cards, this Graphics Card Hierarchy Chart helps to compare the AMD/ATI Radeons with Nvidia GeForces (and Intel integrated video). Even if the video cards are on the same tier, they're NOT exactly equivalent to each other, but they're roughly close for the purposes of this chart. Unfortunately there are no details on the performance differences between each tier, so this chart is mostly useful for comparing equivalent video cards between the major manufacturers. A lot of these video cards are no longer being sold new, but the chart does give you a general idea where your video card may stack up compared to others. For CMBN I would suggest at least being at the level of the following tier or higher:

GeForce 6800 Ultra, 7600GT, 7800GS, 8600GS, 8600GT (DDR3), 9500GT (DDR2)

or

Radeon X800 XT/PE, X850XT/PE, X1650XT, X1800GTO, HD 2600XT, HD 3650 (DDR3), HD 3670

CMBN WILL RUN WITH SLOWER VIDEO CARDS, but there will possibly be sacrifices in terms of model and texture quality, etc. Nothing is quite set in stone and users have different ideas of what is and isn't acceptable in terms of video performance. Large battles with larger maps and larger numbers of units might not be fully 'fluid' to all expectations, even with a fast system. CMBN is still in BETA and thus it is not fully finished to help determine what type of performance you will get with particular video cards. Again, CMSF should be a somewhat accurate gauge as to how well CMBN will play, with the exception that there will be more and larger trees, more detailed buildings and possibly more textures overall with the game.

Another caveat of the above chart is that it may not be exactly descriptive of the differences in performance when it comes to CMBN. The water effects are not going to be of the same quality and features as a number of FPS games, but they'll be above what was in the CMx1 series.

I think this would help a lot with Vista 32bit

At a command prompt (might need admin rights, or UAC disabled)

Code:

BCDEDIT /set increaseuserva 2900

or

BCDEDIT /set increaseuserva 3072

[_Itchy]

I am going to purchase a new laptop soon with the primary purpose of playing CMBN and CMx1 titles. From the benchmarks it looks like Crossfire ATI 5870M beat Nvidia 480M. Are there any known issues with either card and CMx1 or CMx2?

[guachi]

I find it odd that CM:BN won't use more than one core considering multi-core processors have been out for PCs since mid-2005. It would seem that using them would speed up the turn calculations.

[dan/california]

I think they are holding out to do multi-core and 64 bit in one huge rewrite. Given that 64 bit does not have any thing like complete market penetration, and it will probably take several months of programming time to sort out, you can't blame them for waiting as long as they can. And that means no other game improvements for most of those months, remember.

In my opinion may be best to wait until the ability to offload non-graphics functions to the graphics card becomes more common. Given the programming and memory management hassles in getting two cores to work on the same thing it might or might not be worth the effort. Offloading things like LOS calculations to the graphics side whereit's they can run on a 64 or 128 cores is, however, would result in a massive speed increase. But this possibility has just recently peaked out of the development labs and is some years from widespread adoption. A game company has to wait for fairly widespread implementation of a new trick like this before they design the whole game to work around it.

[Erwin]

Since some software tasks benefit from multicore and others do not, how can one be certain that the CM2 software would benefit?

Since it seems to be still a while off for most mass-market games to benefit from multicore, it would be even longer for a niche game.

My understanding is that assuming you are not multitasking while trying to play a game, multicore is an expensive and unnecessary feature for a new system. Maybe in 5 years...

[SNAFU]

I think they are holding out to do multi-core and 64 bit in one huge rewrite. Given that 64 bit does not have any thing like complete market penetration, and it will probably take several months of programming time to sort out, you can't blame them for waiting as long as they can.

In my opinion may be best to wait until the ability to offload non-graphics functions to the graphics card becomes more common. Given the programming and memory management hassles in getting two cores to work on the same thing it might or might not be worth the effort.

A game company has to wait for fairly widespread implementation of a new trick like this before they design the whole game to work around it.

Very good points. Making a program multi-threaded can be described as running trains all in one direction. If they are going the same speed on one track it's no problem--just line them up one after the other. But when you're trying to merge 2, 3, 4 (or more) tracks all leading to the same destination it can be very tricky to avoid collisions. One minor bump and the whole line is jammed ie--the program crashes. I certainly understand why CMBN would be written as a single-threaded application. It saves a lot of programming time, avoids potential pitfalls with future hardware/software changes and keeping it 32 bit means it'll work on the widest range of existing systems. All this equals stability and fewer nasty forum posts about the program crashing when the air-support arrives

Past experience with BF programs would lead me to believe that a video card of moderate power will handle the game fine. I'd be leaning more towards a new monitor (well--actually I just upgraded to a 23" Samsung). As far as CPU's go, the multi-cores are better than single cores even with single-threaded, 32 bit programs. If considering a CPU upgrade, definitely go multi-core. Prices are great right now especially with Intels release of its new Sandy Bridge line. The older models fall in price and AMD's are a fine alternative as well. I would imagine a good CPU will make a tremendous difference with the new sighting/LOS system...

I won't presume to claim what hardware requirements will actually be but I'm betting a bottle of Leininkugel's Amber they'll be very reasonable. No doubt those stats will be revealed before the actual release of the game.

SNAFU

[moneymaxx]

Since some software tasks benefit from multicore and others do not, how can one be certain that the CM2 software would benefit?

There are calculations in the game that effect every unit. Take LOS/LOF, TacAI, pathfinding for example. I dare to say, obviously without inside knowledge, that those use quite a lot of calculating time. On a single CPU you calculate for unit 1, then 2 and so on. Now imagine you had 100 cores, 1 core for every unit calculating not sequentially, but at the same time in paralell. In theory much faster.

[guachi]

I can understand waiting a bit for 64-bit support - according to the Steam survey, about 50% of PCs are 64-bit and 50% are 32-bit.

However, 90% of PCs have more than one core. I have no idea how easy/difficult it is to program for, but it's something that would be beneficial to almost all the customers. I think the only single-core processors currently available from Intel and AMD are their ultra low power CPUS.

[YankeeDog]

Steve has made statements in the past to the effect that the increased performance from re-writing the code to take advantage of multicore support would be minor, and that the coding work would be considerable. Basically, small benefit for a lot of work. So for now, they're spending their (limited) coding time budget elsewhere.

I'm sure it will happen eventually, but probably not for the CMBN family.

[finalcut]

I am planning a major computer upgrade myself.It would be cool if you guys would let us know when the game is coming out so I can get to work on my upgrade.

[Redwolf]

Doing a 64 bit version is trivial compared to making the thing multithreaded.

A 64 bit version of CM is also almost completely useless to the user.

In any case, don't plan on threads anytime soon. If you want CMBN to run fast, get something with a high per-clock speed.

[dan/california]

Doing a 64 bit version is trivial compared to making the thing multithreaded.

A 64 bit version of CM is also almost completely useless to the user.

In any case, don't plan on threads anytime soon. If you want CMBN to run fast, get something with a high per-clock speed.

Honest question, wouldn't 64 bit allow the use more memory, and therefore larger maps and scenarios? Or does the CMX2 engine consistently hit some other limiting factor first?

[gibsonm]

Perhaps but then you need two versions a 32 bit one and a 64 bit one.

AIUI you can run a 32 Bit app in a 64 Bit OS but not the other way around.

[Redwolf]

Honest question, wouldn't 64 bit allow the use more memory, and therefore larger maps and scenarios? Or does the CMX2 engine consistently hit some other limiting factor first?

In theory, yes. In practice, no.

BFC would not drop the 32 bit client running on a 32 bit OS. So they have to write the code in a way that at any point in time less than 3 GB of virtual memory mappings are in effect. So even if you had a 64 bit client it would still map less than 3 GB, given the code doing the same.

This doesn't mean that less than 3 GB of RAM is used on machine that have more OS-visible RAM. The limit only affects concurrent virtual memory mappings. You can change mappings around and you don't have to map. If you use file I/O then the OS which has more RAM will still hold all the content in RAM even though things might not be currently mapped in the application.

Also, I highly doubt that virtual memory is the currently limiting factor for map or scenario size. It is much more likely that general performance concerns (name exponential increases in required computing for LOS and the like) are the limiting factor.