Renderosity Forums / Poser - OFFICIAL
Welcome to the Poser - OFFICIAL Forum
Forum Coordinators: RedPhantom
Poser - OFFICIAL F.A.Q (Last Updated: 2025 Feb 03 12:46 am)
Subject: What kind of CPU works best with POSER ???
udgang, I STRONGLY recommend B, based on evidence, but here is an important question: does that AMD have 1024 L2 Cache? If it only has 512K that is a negative...from recent threads here, that L2 cache is really important. Your HD setup is important too, more important than you think, because Poser goes out to the page file a lot, even if you have a lot of RAM. Video card is NOT a factor in Poser 5 or lower, but makes a difference in 'working mode' (not in render speed) in Poser6 ::::: Opera :::::
speed and configuration of hard disks, are they at least SATA? Suggest you find out about L2 Cache on AMD. The page file is what used to be called the swap file. It's the file dedicated to virtual memory. This is important for Poser optimization, some people put the page file on a completely separate HD than the exe and/or runtime, so it can read/write at the same time other disk functions are executing. ::::: Opera :::::
I've got an AMD L2 cache machine that worked very well in some recent tests we all tried here in the forums. There was a P5 image that was created by Jim Burton (I believe) that we all loaded up and rendered. We timed how long it took to render and the AMD-L2s kicked butt. My computer is about a year old now and it's running smoothly and perfectly and I get really fast renders which I'm still not used to (On my old Pentium machine, I could get up and feed the dogs before it got done rendering but not any more with the AMD L2). Kristta
B: That AMD64 3700, is it a Socket 754 or a Socket 939? Socket 939 supports dual channel RAM (and that means FSB800). And a Socket 939 mainboard supports upgrading to dual-core AMD64 CPUs (expected in september 2005). I'd definitely go for an AMD. Not only do they perform better with Poser than the P4, they're also ready for the new 64 bit operating systems. An AMD64 running Windows XP is effectively running in emulation mode, and still its performance is comparable to a P4 in the same price range. Imagine what happens when Windows XP 64 bit becomes available (this spring!) The beta seems to be very solid, 32 bit applications can get a full 4 GB of RAM to play with, and they're at least as fast as on the standard XP. All AMD64 chips have L2 cache. The difference is in size: some have 512 Kb, others have 1024 Kb. More is better.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
A word about the P4 64bit: it's not 64bit. If I remember right, it's 40 bits, and works with bit shifting. The AMD64 is a true 64 bit CPU.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
Drive speed is very important to Poser. Drive speed as a lot to do with factors outside the HD. In a dream system you would build a RAID5 array connected to a PCI Express controller (And of course a PCI Express board). This would give you the maximum drive through put. BUT a drive array like that costs a load of dough. To cheat for speed you could try a simple stripe array (RAID0) Some main boards with SATA support this directly. The extra dough you're gouing to spend to make the system 64 bit would probably be better invested in a rocklin drive array or more RAM. Poser is not 64 bit so half the CPU's horse power can't even be used on it. There are precious few 64 bit apps out there. Technology rule #1 Never buy technology before you are going to use it. By the time you do use it it will cost half as much! Nerd3D
nerd: good point. I never regretted buying WD Raptor drives - they're FAST! But they're expensive. My machine is an Athlon64 3500+. I chose not to buy a 3800+ and spend the $400 I saved on those Raptors.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
I actually have System A, and the first thing I noticed is that everytime I render, I'm using the maximum amount of ram I have, 1024 Mb (or very close to it). I have my page file/virtual ram set up on my D drive which is a SATA hard drive, that seems to help a little. My C drive which has Poser on it is just an IDE drive. As soon as I can afford it, I'm going to get another gig of ram.
Attached Link: http://www.newegg.com/
At link above, type in 'raptor'the smaller raptors (37Gig x 2) can be had for about $125 each. You can Raid-0 them for a 72 Gig 'working' volume that is rockin'.
::::: Opera :::::
You know, I wonder what type of performance boost you would see with solid-state hard drives. It would have to be phenomenal. Too bad they are so expensive. You probably wouldn't even need to have allot. Like 4 - 6 gigs just for the page file. (Sorry just a thought I don't mean to hijack the thread)
not at all....it has occurred to me lately that an optimal setup might...and only might...be some small stupendously fast drives on board, one for the OS, one for the swap file, one for the poser exe, one for the runtime folder, and then the pz3 on the raid-0 array. Now that's an exaggeration, but I am with you on the idea. I even floated the idea of RAM drives, but others told me that was an idea whose time has passed. In general, I find it amazing that some sort of hardware solution to house the OS and/or swap file on a 8 Gig no moving parts drive has not emerged. ::::: Opera :::::
System B with the "Winchester" CPU (if it's out yet). The 90nm Winchester uses less power and is better performing than the other Athlon 64s. Intel systems, especially "Prescotts" use way too much power and don't perform as well for the price. The best way to increase render performance is to increase memory. Disk speed is secondary to this. Stay away from Raid 0 unless you have good backup capability,
Opera,
The solid state drives exist, however they cost 10 times the amount of a standard drive of the same capacity. (If I remember the multiplier correctly. It could be 100 x but I'm not sure. I changed professions 1.5 years ago and I have not been in the IT mainstream.)
The drives are fast though. You click on an app and its running. Man I would love one of those. Hell if you ran the OS on a drive like that it would scream. Imagine pushing the start button on the computer and being @ the desktop ready to work in 1-2 sec!
I can see this technology really being put to use on laptops. Lower power consumption and faster data throughput.
Well I guess I will just be telling my grandkids what an Apple ][e was like.
Message edited on: 03/29/2005 20:31
"The best way to increase render performance is to increase memory. Disk speed is secondary to this. " Once you are at about 1 - 1.5 Gig, this is not true. "Stay away from Raid 0 unless you have good backup capability, " This is dubious advice. Nothing wrong with RAID-0 with regard to backup. You need the SAME backup policy with ANY primary drive. In fact, a case could be made that a RAID-0 array with superior drives is MORE RELIABLE than a cheaper-made single drive system. ::::: Opera :::::
I am running 1Gb of RAM. The disk is not an issue in any of my renders, unless I'm multi-tasking with another program. Of, course I don't render scenes with lots of characters like SVDL does. He certainly pushes Poser to the extreme. I don't know what OperaGuy renders. The swap file is virtual memory. When you "use up" your physiscal RAM, the "least recently used" data/program is swapped out to virtual memory to free up physical RAM. If you are thrashing your swap file, then you need more RAM. Most home users are lax about backup. This makes Raid 0 more of a risk. With Raid 0, you double your risk of data loss due to drive failure (compared to a single drive). With Raid 1, you risk of data loss is halved, both drives have to fail in order for you to lose data.
But Raid 1 does not speed you up. Better to teach home users about proper backup REGARDLESS OF CONFIG than advise them that a totally valid solution to Poser speed is bad. "I am running 1Gb of RAM. The disk is not an issue in any of my renders, unless I'm multi-tasking with another program." Either this is theoretical, or you are successful in your setup when others are not. All the threads I have participated in...no matter how much RAM you have, Poser STILL swaps out to virtual memory A LOT no matter what you do. This is infuriating and counter-intuitive; one would think that if you put up a lot of RAM, Poser would stay in RAM until pegged out. But this is not the case. I render quite low density scenes, like one naked low-res EJ and complex hair with light props in 3-minute renders. This is for animation. I have 4Gig Ram. My RAID-0 array thrashes the whole time, and watching in the task master window, I am pushing RAM usage up to only about 800MB. If you can teach me how to make Poser NOT THRASH, but rather make use of another 1.2 GIG RAM, I would be grateful. ::::: Opera ::::: AMD Athlon 64 3500+ 4Gig RAM 2 x 37G Raptor in RAID-0 as working director 250 External Firewire drive, primary storage, backed up with DVD including frequent off-site. XP Pro SR2
For a HOME systems backups are critical for Raid 0 configurations. They are not as critical for Raid 1 configurations, since the data is backed up continuously "on the fly." There are things that are more important than performance. One of these is fault-tolerance or "Up-TIme". If a primary disk in a Raid 0 configuration fails, you will have "Down-Time" and data loss. If a primary disk in a Raid 1 configuration fails, you will still be able to use your system with no data loss. You better do something about the bad disk, though. No one has ever shown me a Poser benchmark that shows that significant performance gains can be attributed to a Raid 0 configuration. Benchmarks done by Anandtech and Storage Review indicate that there is no "real-world" gain in performance. Performance is limited by PCI Bus Bandwidth. PCI-Express may be a solution, but right now it's only for Video Cards. For starters, try disabling services you don't need. For exampel, if you don't share your files and printer, then disable Server Service and NetBios Helper. I wonder if the AntiVirus package makes a difference. I'm using McAfee. What are you using. Maybe I'll try some renders with it disabled.
Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=2077843
nakamuram. RE: Poser render Speed >>Benchmarks done by Anandtech and Storage Review indicate that there is no "real-world" gain in performance.<< This cannot have been a Poser-specific test. Poser has it's own strategy, which must be respected. I have been involved in several threads that delved deeply into this including runs of the Poser-specific Jim Burton benchmark (see link attached), and here is my general conclusion (Poser6 not explored for this conclusion): No matter how much RAM you have, Poser 'thrashes the swap file' and loads textures from the drive. There is no such thing as sending a Poser pz3 off to render and have completely silent (except for the tiny tiny moment it actually writes the tiny tiny render file to disk) hard drive during render. Once agin, if I am wrong about this, please teach myself and many others who would be very happy how to force Poser to stay in RAM during render when enough RAM is present. Therefore, to get great render times, you have to have blazing Hard Disk management. First, the disks themselves must be fast, at least SATA and preferably the 10,000 RPM Raptors. Next, you have to manage the 'five elements' with regard to HDspeed: OS, OSSwapFile, Poser.exe, RuntimeFolder, Pz3. If you have one modest harddrive and everything is on this drive, you are condeming your renders to slowdown. The first line of attack would be to at least to defrag and partition that drive and move the swapfile to its own partition. But you still can only do one read/write at a time. The HOME or hobbyist setup? Get your RAM to 1G or 1.5G, then if you have only one HD to your name, anywhere, consider investing in a low-cost second small SATA drive. The next level of strategy is to place different parts of the five elements on different Hard Drives. I do not claim to know the ultimate best way to do this, or what amounts to overkill, etc.; I am still experimenting. Maybe the most radical approach would be five separate drives with dedicated controllers. Solid state (RAM hardware drive) is way radical, but interesting. At any rate, one strategy is to RAID-0 two Raptors. You get blazing individual HD speed, plus the software makes your 'where should I put this' decisions for you. Simple, fast, efficient. These drives cost about $130 eash. Several of my poser mentors have gone this way, as have I. slvd and I have nearly identical rigs, and our Jim Burton P5 benchmark stands at 158. That is a great time. Please also see post 12 by nerd above. Although this thread asked about 'best CPU' we have answered that but also supplied deeper information to the original poster (and others researching in the future) about the critical necessity for fast HD in Poser. ::::: Opera :::::
Attached Link: http://www.renderosity.com/messages.ez?Form.ShowMessage=2130434
The "OTHER' element of great Poser Render speed: L2 Cache size. While I am thrilled with my rig and it's 159 Jim Burton, there are faster times being reported. The difference, in my opinion, is that some AMD CPUs have 1024 L2 while mine has only 512. Please see attached link for discussion. ::::: Opera :::::
Y'all will hate me...but my advice wold be to get a Mac..all my Poser probs have disappeared since I moved it from the PC to the Mac, except for some interface quirks. It is however a G5 with 3 gigs of ram. No crashes on renders, Firefly gives me no problems, all my "pc only" content loads right up (sometimes have to convert the thumbs is all), no crashes during cancels or figure changes, and I can load content on the fly without restarting Poser. I can also have a HUGE Photoshop files open (we're talking over 400 MB) and about 6 other apps as well and switch effortlessly between them even WHILE rending a 8x11 300 dpi image. Every time I get on a PC I immediately want to tear my hair out, even in Word. Just my 2 cents...
Rendo Store | Freebies | RDNA Store
I have a G5 with 2.5 Gig. I use it for music, Logic7. I do have a seat of Poser5 for it and have done some rendering...the speed is not there for me, although since I have it, I do use it for Poser sometimes when I need to set a render going on all my machines overnight. I WILL be upgrading my MacPoser5 to 6. ::::: Opera :::::
Opera: In Jim Burton's Benchmark, my machine (with Raid 1) is the best in it's class (P4 2.8). It's faster than several P4's with higher clock speed and Raid 0. It's not an A64, but its not a Mac, either. Granted that the Anandtech and Storage Review Benchmarks were not Poser-specific, but they were carried out scientifically and objectively. Why don't you (or some of the other Raid 0 advocates) show us some benchmarks to support your claim that Raid 0 adds a significant increase to Poser's rendering performance?
A very interesting fact is that the Macs were MUCH slower in the Jim Burton benchmark that one would expect. A dual G4 should run circles around a midrange P4, but it doesn't when it comes to P5 rendering. Apparently the Athlon64 is the fastest CPU for Poser renders, especially when it has 1 Mb L2 cache. About RAID0: it is not a complete valid comparison, but my AthlonXP 2700 / 1 GB DDR333 /2xATA100 RAID 0 outperforms my P4 2.8 / 1.5 GB DDR400 Dual Channel /1xATA133. Not only on Poser, also on most other apps. Only games run better on that P4, but that's due to the graphics card, a Radeon 9600Pro is significantly faster than a Ti4200. I suspect the RAID 0 setup is part of the reason that the old AthlonXP rig performs better than the P4. I could break the RAID setup and run it as two separate disks, but then I'd have to reinstall all my stuff. Since I'm using that AthlonXP for a lot of other applications, this would take far too much time. It runs fine, fast and stable as it is (the machine is 2 1/2 years old and it's almost always on). I'm not going to change the machine.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
Winchester CPU? That is a new one on me, I do remember when you used to call Hard-Drives Winchester drives, though. I even know where the name came from, it had to do with being on an IBM 3030 (30-30), and if you aren't a gun nut you would never know what I'm talking about. ;-) Anyway, the P4 and the P5 CPU tests are in my Freestuff here, and both have a list of results for various computers, I think, as do several of the threads on that subject in this forum.
SVDL, I don't think you can attribute the performance difference in your systems to disk configuration, especially since the Hard Disk is not a factor in Jim Burton's Benchmark. My P4 2.8C (192 secs) is faster than both of your machines (Athlon XP -- 205 secs, P4 2.8 -- 210 secs). We need to find two people with the same machine, except for the disk configuration, then run Jim Burton's benchmark to establish a baseline, then run a benchmark that works the disk.
nakamuram: "In Jim Burton's Benchmark, my machine (with Raid 1) is the best in it's class (P4 2.8). It's faster than several P4's with higher clock speed and Raid 0 Where did you find the information on the other P4's with RAID-0 to make the comparison? "My P4 2.8C (192 secs) is faster than both of your machines (Athlon XP -- 205 secs, P4 2.8 -- 210 secs)." svdl and I are getting 159 on the Burton test with our RAID-0 systems as compared to your 192. and then you say this: "the Hard Disk is not a factor in Jim Burton's Benchmark" I don't understand that. Yes, when reporting their Burton results, most people do not report their HD setup. But per my several posts in this thread about how Poser swaps, how can you say the HD is not a factor? And like Steven, my RAID-0 is so successful and to take it down for a comparison would be a gigantic hassle, I will not be doing the comarison you are interested in. However, upon the purchase of my next computer, I WILL be doing some rigourous HD setup comparisons. ::::: Opera :::::
OperaGuy, Read carefully, please. I am comparing my P4 2.8C to SVDL's Athlon 2.7XP and P4 2.8, NOT his or your A64. I'm looking at upgrading to an A64, so I certainly wouldn't claim my machine is faster. In the Jim Burton Benchmarks, some people stated their disk configuration. I believe that Jim Burton himself had an overclocked P4 2.8 with a Raid 0 Configuration. My machine was faster.
HEY !!! I'm the one who started this thread. Just wanted to let you know that I decided to go with computer B: AMD Athlon 64 - 3700 - 400MHz front side bus Radeon 9550 (256 MB) 1024 Mb ram (400 MHz) ... I wanted to try the Jim Burton-thing, but I can't load it in Poser6 - missing some textures. Anyone want to make a new "benchmark" for Poser6 ??? :)
...and the 1024 L2 Cache... That rig oughtta rock! If you are not going to operate on more than one drive, I suggest at the very least you start with the clean install and partition the drive. The page file should definately be in its own partition. One secondary system I use for poser I have a separate partition for the OS, for the page file, for poser. ::::: Opera :::::
Just on a theoretical basis....
How could Raid-1 be faster than Raid-0?
As I understand it, RAID-1 is for redundancy and safety, right? Using the example of two 37GIG Raptors, with RAID-1, you have total drive yield of 36Gig and the speed is either slower than or equal to that of a single Raptor, since the data must be written twice.
With RAID-0, you receive a yeild of 72GIG and it HAS to be faster, since you can write a given 'save' to two DIFFERENT drives at the same time, not to mention read/write at the same time.
Please correct me if this is inaccurate.
And yes, I understand the risk factors involved. But I am just talking about speed itself -- the dangers of RAID-0 can be dealt with easily in other ways.
::::: Opera :::::
Message edited on: 04/01/2005 07:40
Hard disk performance is definitely a factor in the benchmark - watch the page faults, they represent disk access.
The pen is mightier than the sword. But if you literally want to have some impact, use a typewriter
I do have a Raid 0, and Operaguy has it right as far as I know. If you happen to have a motherboard that supports it and plan on a set of matching drives I think Raid 0 is about the cheapest real improvemt in computer performance you can have. You also don't loose any capacity, compared to the redundant RAIDs. The downside is, of course, that you are doubling the chance of failure over even a single drive, 4X the failure of two regular drives (where you only loose the one drive), an if they go you loose everything. Back up often! Don't buy Western Digital Drives! ;-) I'm up to a Pentium 3.2 now, overclocked to a little over 3.5, and it still isn't as fast as some of these guys, though. (190 seconds) And I've optimised my system every way I know, consistent with reliability.
I've worked with servers professionally. I have never seen a server setup with a Raid 0 array, unless it was part of a Raid 0+1 mirror. On the servers I have setup, I've used Raid 1 and Raid 5, but NEVER Raid 0. Raid 1 halves your available storage, but disk space is cheap nowdays and fault tolerance is wise usage of disk space. You guys who religiously believe in Raid 0, answer these question with hard facts -- 1) "How much performance does one gain by going to a Raid 0 configuration?" 2) Is the performance gain worth the risk in down time and data loss? Put your money where your mouth is and be honest about it!!!
Jim, I am with you totally on twin SCSI, that is the best route, with separate conrollers, right? In the desktop music recording studio world, this is the default, SCSI array in RAID-0, especially for direct to HD high bit rate recording of voice and acoustical instruments. You've GOT to have that speed of double write. Maybe you SHOULD wait for the SR. Steve from CL just posted in another thread that it is almost ready. If you put up a benchmark now, people will start running it with the P6 memory problem in place and confuse the issue. It would not be a fair measure of P6. Just my opinion. ::::: Opera :::::
nakamuram,
You keep avoiding the point. We are not installing networks and database workgroups, etc. So your points about outside benchmarks and how you never see RAID-0 in your work are perfectly true and valid, but in the context of Poser on a single CPU are irrelevant.
This is a special case, special purpose deployment or RAID-0
Here is a summary of the argument, once again.
- Poser thrashes the swap file.
- You COULD run on a single HD, perhaps with a protected partition for the swap file, but that is pretty slow.
- RAID-0 gives simultaneous write/read to two fast drives.
So in answer to your "1) "How much performance does one gain by going to a Raid 0 configuration?"
The answer is
a) in comparison to what?
b) to my knowlege no one has done a fair TRUE oranges to oranges measurement of RAID-0 against RAID-1, nor RAID-0 against two drives with no special config nor RAID-0 against a one drive setup.
But....we DO KNOW IT IS FASTER than all of the above. We have plunked down hard cash to get this advantage. As I stated above, I myself am still experimenting. When I build my next system, I will take time to do some real hard-assed comps.
Now, on your second point. Anyone advanced enough to be installing RAID-0 is, or should be, informed enough to know the simple, direct truth that if one drive fails, your data on the R-Volume is gone.
Here's how I think about this, and I offer it as a suggestion to all serious computer users, take it or leave it.
CONSIDER YOUR WORKING HARD DRIVE ALREADY DEAD.
In other words, install all your apps and the OS on a drive which you might ghost, or not, as long as you put your install disks in a safe off-site location. If that drive fails, just re-install all from original disks or the ghost.
Next, the directory(s) in which you run your object files, your database tables, your word docs, your downloaded iTunes files, and yes your precious Poser pz3 files, make it as fast as possible, but consder it dead. This is true if it is a one-drive 80Gig IDE hard drive, or a RAID array of any sort.
So, create an effective, usable, rigourous backup of your working volume, including off-site storage of media on a regular basis.
Now that last point....I submit that RAID-1 is DANGEROUS from one point of view. It could lull you into a false sense of security. "Oh, my working volume is very safe because of the RAID-1 redundancy, so I can sleep well knowing if I DO have a HD failure, my volume is intact."
This might lead to a weak off-site backup habit. And massive grief on the day someone steals the system, or a serious accident destoys both HD.
So, anyone serious about data is backing up, period, and taking offsite. This means: RAID-0 is perfectly safe.
::::: Opera ::::
Message edited on: 04/01/2005 23:45
Privacy Notice
This site uses cookies to deliver the best experience. Our own cookies make user accounts and other features possible. Third-party cookies are used to display relevant ads and to analyze how Renderosity is used. By using our site, you acknowledge that you have read and understood our Terms of Service, including our Cookie Policy and our Privacy Policy.
I'm going to buy a new computer, but am not sure what to get. I have 2 different computers to choos from: A: Pentium 4 (64 bit) 3.2GHz - 800MHz front side bus. Radeon X740 (128 MB - 900 MHz) 1024 Mb ram (400 MHz) B: AMD Athlon 64 - 3700 - 400MHz front side bus Radeon 9550 (256 MB) 1024 Mb ram (400 MHz) ... do anyone know which might work best/fastest with POSER 6 ?!?!?!