Peter I have tried using different colors offline and after about a month, it starts to look really bad. I think what I am going to do is at the end of the article I have put a section with the date and the update made to the main article.

Right now the article is pretty much complete with the exception of new driver updates and new video cards to add to the list, so I think my idea of putting the update info at the end would work. At least I hope so.... lol.

Thank you, Harm and Bill, for your service. I am trying to get some insight into optimizing for CS5 performance thanks to your benchmarks.

I was having problems trying to just copy and paste from the benchmark results page into excel as it would be tab-deliminated data and wouldn't import properly into excel. However, I was able to get the data into Excel via a web query.

Just being able to sort through and filter things makes it easier to analyze the data and draw conclusions.

Anyways, I already appreciate all the work you've done putting the tests and site together and updating them.

Thanks again,
Peter

I noticed that some of the faster Disk Access setups also have faster non-MPE speeds. Most of the faster setups have RAID0 setups, some with SSD drives and some with up to an 8 drive RAID0! Is this even safe? Wouldn't it be disastrous if even 1 drive were to go bad, especially in BillG's 8 drive RAID0 setup?

If even one drive in a RAID 0 setup goes bad, the data (if any) on the other seven becomes unusable. (At least without the need to use astronomically expensive data recovery services.)

Though I have to admit that even single-drive setups are not immune to total data loss. In fact, the very slowest Disk Access setups are those systems that use just a single drive (or single drive volume) for everything - the OS, programs, projects, renders, page file and video source and output files. Those are the systems that have the greatest risk for data loss due to the extra wear and tear on the single drive. Those are the very reasons why Adobe does not recommend single-drive systems at all with any of its prosumer-level Creative Suite applications.

And yes, a two-drive software RAID 0 array on an Intel SATA controller does not improve the disk access scores enough over a very fast single disk (separate from the OS disk) to justify the trouble of creating such an array.

Peter,

This is a very valid and good point you make. Normally no one would even consider a 8 disk raid0 for editing, but for benchmarking it is OK, at least for that purpose only, IMO.

Rendering does require that the preview files are written to disk and then disk speed does come into the picture as well, not only CPU/GPU/memory speed.

Bill is constantly trying out various disk setups, with multi SSD's, SCSI's, SATA's with different cards (LSI, ARECA 1680, and shortly 1880) and different raid configurations. I always wonder how his basement looks with all the stuff he gathered in the past years.

It does make for interesting scenarios just to see "what if?" However, as you mention, it is not safe in a real world application to have crazy RAID0 setups. I can't imagine trying to back up your 8 TB setup in case you do get a disk failure...

I think performance can also be tweaked by specifying the speed and latency of the RAM used. From articles I have read, the latency improves performance more than the speed of the RAM. I wonder if this has any effect on the benchmarks since we know the amount of RAM available definitely makes a difference.

This is not the case, based on my testing. In fact, memory latency ratings have very little impact on the CS5 performance. What's more, CL6 or CL7 memory @ DDR3-1600 speed might actually perform slower than CL9 or CL10 memory at the same memory speed unless there is crazy cooling on the CPU because such ultra-low-latency memory typically requires a much higher DIMM voltage than standard DDR3 memory at the same memory clock speed. And this is all because raising the DIMM voltage also increases the voltage of the memory controller, which in turn requires a raising of all of the other CPU voltages as well in order to minimize damage - and all of that raises the temperature of the CPU itself significantly. With some CL6 or CL7 1600-speed memory and weaker IMCs, the CPU might start throttling back at even stock CPU speeds. That might have forced the user to severely underclock the CPU in order to keep temps under control.

Really? I thought a RAID0 gives 2x the speed? The only benchmarks I could compare were from Dieter's_PC and Dieter's_PC_upd. He changed to a RAID0 and his MPEG2-DVD was almost a minute faster from 153s down to 98s.

So, just get the cheapest RAM available? That would be great news :)

Peter,

Keep in mind that there is a difference between disk access time and transfer rate.

In general creating a raid0 does hardly anything to disk access, it remains approximately the same or slightly faster, but that difference is negligent. What does change is the CPU load, especially with software raids, that increases a bit but again not enough to be worth talking about and - this is where the advantage is - it increases the transfer rate of the data almost linearly. A 2 disk raid0 is nearly twice as fast as a single disk, a three disk raid0 is nearly three times faster than a single disk, etc. Again not in disk access but in transfer rate.

A modern day disk can achieve transfer rates of around 100 MB/s, a 2 disk raid0 can achieve almost 200 MB/s but in both cases the disk access will be around 13 ms (based on 7200 RPM disks).

One caveat: Do not expect that this linear behavior continues indefinitely. The bandwidth of the PCIe bus will at some point become the bottleneck. On my array (12 disk raid30) I can't get better performance than average 853 MB/s transfer with a disk access of 10.7 ms (Areca based). A single disk comes out with about 100 MB/s transfer and 13.5 ms access and a two disk raid0 comes out with around 180 MB/s and 15.8 ms access (Marvell based). BTW this is all based on the old Samsung F1 disks and the newer F3's are a lot faster.

Please, could you explain the difference between disk access time and transfer rate in terms of Premiere CS5 usage? How are the two relevant in the benchmark testing?

Thanks!

On my current system, I could not get better performance than a 92-second result in the AVI Disk Test portion of PPBM5 with my particular two-disk RAID 0 array. And that is with two 1TB Seagate 7200.12 hard drives. I will re-run the test with one of the drives disabled and the other in regular (JBOD) mode, and check back later.

Peter,

Consider the situation you need to look up a phone number in the phone book or yellow pages. Access time is the time it takes you to get to the relevant page. Even if there were a number of people looking for that same name and number, it would still take them about the same time to get to the correct page in the phone book. That can be called access time.

Now consider that you need all the numbers from a business or government agency, all on the same page and this may amount to 20 or more numbers. Whether you are alone (single disk) or with a colleague (2 disk raid0) makes no difference in access time. You both need around the same time to find the required page (access time). But now, once you both found the page you need, things get easy if you are working together (raid0). You can write down the numbers, one doing the even entries, your colleague the uneven ones. That means you get the job done in half the time compared to if you were alone. That is transfer rate.

Hope this explains it a bit.

I just got done retesting my main system with a non-RAID disk, and found that the AVI Disk time in PPBM5 increased by about 40 percent -- from 92 seconds in a 2-disk RAID 0 array to 129 seconds as a single drive.

Now I discovered why my RAID 0 array was performing significantly slower than expected: Although the two drives were both Seagate 7200.12 1TB drives, one has a slightly different firmware revision than the other. Tested individually, the one with firmware version CC37 had an access time of nearly 16 ms while the one with firmware revision CC38 had an access time of only 13.6 ms. Put those two together and I ended up with noticeably slower RAID 0 performance than expected.

But back to the graphics cards:

Although a GT 240 with 1GB of DDR5 memory can suffice for CS5, many such cards include barely adequate coolers. A card with good cooling is strongly recommended for CS5. Unfortunately, the best coolers are usually found on premium cards that the manufacturer claims can be overclocked (or have already been factory overclocked to a significant degree before shipment).

Was wondering if you guys have any opinions on Mushkin Enhanced Silverline RAM. It's on sale for $175 for DDR3 1333 12GB (3x4GB) after rebate. I couldn't find many reviews on it so I thought I'd ask here.

Link: http://www.newegg.com/Product/Produc...82E16820226096

Thanks

Peter,

I have no experience with them. Looking at the price it sounds attractive, but the 9-9-9-24 timings disappoint me for 1333 memory and that may be the reason for the relatively low price. Had it been 1600 memory with these timings and price, I would have said: "Order two sets for me too. You may get extra rebate".

Generally, yes, you do get what you pay for. But tighter latency timings sometimes hide the fact that such memory requires a significantly elevated DIMM voltage just to even run at normal JEDEC timings and the slowest supported speed. I have run into early low-latency DDR3 modules that required a very high DIMM voltage of 1.9V just to even run at stock speed (and that's not to mention that these particular modules do not like to be run at even 1.65V, let alone 1.5V, at any speed or timings). Couple this with the normally low operating voltages of the rest of the CPU, and high-voltage DIMMs can (and do) significantly shorten the usable life of a CPU. And such failure is not warrantied at all by the CPU manufacturer.

So what do you think about the Mushkin RAM deal, Randall?

Thanks

Not as good as you might think. Remember, rebates are never a sure thing especially since you might not even receive the rebate check at all (lost in the mail); Mushkin is one of the few tech companies whose third-party rebate administrator actually sends you a check rather than a prepaid Visa debit card. So, not counting the rebate check that you may or may not receive, that Mushkin RAM actually costs you $205, not $175. That's not that good of a deal considering that it is priced too close to a 9-9-9-24 DDR3-1600 kit for comfort.

Though the Mushkin kit is not that bad since at least it is rated at 1.5V, and 9-9-9 is at the loose end of the official JEDEC latency standards for 1333-speed DDR3 memory. (Remember, the official JEDEC standard CL range for DDR3-1333 memory is CL7, CL8 or CL9.) Some other inexpensive kits of similar capacity, price and speed require a significantly higher than standard voltage just to work reliably at even the lower JEDEC speeds and looser timings.

So if this RAM was DDR3 1600 with 9-9-9-27 timing, would it be a great deal? I'm looking at this Kingston RAM for $90 after $30 MIR. So it would be $180 for 12GB.

Thanks for your inputs!

Re: Best Graphics Card for CS5
 

Hello All,

First, my many thanks to Bill Gehrke and Harm Millaard for compiling an excellent and exhaustive list on CS5 systems vs. nVidia video card performance. You both are truly a credit to the community.

So. I've finally upgraded (better late than never) to Production Suite CS5. My system: I'm running a Dell i7 920 435MT (2.67GHz stock), 12 Gigs of RAM (running @ 1066MHz) and Windows Ultimate 64, three physical disks (no RAID); two internal (SATA) and one external (USB 2.0).

For overclocking, I've had to implement a software overclock, to 2.91GHz, as the BIOS of the 920 will not permit overclocking and there are no BIOS upgrades for that model that I've found. Basically the Dell 920 is an "entry" level i7, thus Dell seems to cripple them right our of the gate.

So now, decisions, decisions. According to Bill and Harm's spreadsheet, it would appear that the GTS 450 might be the best choice for CUDA enabled MPE rendering. I was pretty much set on that card. But some in this thread have mentioned the GTX 460.

Going off David Knarr's information at Studio1Productions.com, (thank you, David), the GTS 450 has:

192 CUDA Cores
Standard DDR5 Memory (1024MB Typical)
128 Bit Memory Interace Width
783 Mhz Core Clock

The GTX 460:

336 CUDA Cores
Standard DDR5 Memory (1024MB Typical)
256 Bit Memory Interface Width
675MHz Core Clock

Since I wanted to stay around $200 or under, I guess it would seem obvious that the GTX 460 would seem to be the most reasonable choice. But looking at the results of MS Studio1's system in the spreadsheet (an i7 920 system similar to my own, albeit overclocked to 3.3GHz), it looks as though the GTS 450 fared very well in the tests, at least relative to other systems of the same caliber.

I've already come to grips with the fact that I'll be upgrading the power supply (Corsair 650W), so there'll be plenty of power for either card. (I also plan to increase to 24 Gigs RAM). So what do you think, guys? 450 or 460? I've also considered, as we all always have to do, that certainly Adobe will mostly likely upgrade and make better use of the card and CUDA in later (CS6?) versions of Premiere, and perhaps even After Effects. So how 'bout it guys? All opinions welcome.

Rob

Re: Best Graphics Card for CS5
 

From my own testing, I found the gfx card to have more of an effect only with more layers and several effects on each layer. And when MRQ is enabled, the gfx card is involved even less. Harm and Randall posted their GPU usage during MRQ rendering and it was very low; thus, the CPU is the most important followed by ram.

I would just go with the GTS450. Save the money and get another drive, Seagate or Samsung or a SSD for the OS. I have an Intel X25 80GB in my HP and also a Seagate 7200.12 160GB in the same PC with Win 7 installed on both. The SSD is for 'work' and the Seagate is for testing and Matrox. I recently used the Seagate OS and WOWWWW is everything so much slower than the SSD. Premiere takes 3-4 times as long to open and same with AE. Trying to open a program and do something else on the Seagate takes far longer.

ONCE YOU GO SSD, YOU WILL NEVER GO BACK!

Re: Best Graphics Card for CS5
 

Steve, thank you for the heads up on the SSD situation. I had looked into it in the past but had dismissed it. I must confess your mention of it has rekindled my interest.

Currently I'm looking at the OWC 120GB Mercury Extreme Pro and the Intel X25M 120GB. Apparently the OWC has some sort of better management of wear leveling over time, if you can believe the adverts. And based on what I've seen on YouTube vids, seems to the remarkably fast. Heck, maybe I'll just get three; one for the OS (clean install Win 7), one for the project/read drive and one for the export drive. I think I could be sure my drives will not be the cause of a bottleneck after that. No, after that it will be my DDR3 RAM which is only running at 1066MHz. But conventional computer wisdom says there has to be at least one bottleneck somewhere.

On the card, I'm probably going to go with the GT 240 (1GB DDR5). With the MPE in the current version of Premiere CS5, it seems to perform about at well as every other card.

Re: Best Graphics Card for CS5
 

My further testing of both the GT 240 with 1GB of DDR5 memory versus the GTX 470 revealed that the GT 240 is definitely slower than the GTX 470 in the PPBM5 tests when both cards are installed in identical i7-9xx systems at the same CPU clock speed (3.67GHz) -- about 10 seconds slower in MPEG-2 DVD encodes and about 20 seconds slower in AVC Blu-ray encodes. However, that difference, although noticeable, is still much smaller than the difference in price between the two would have led someone to believe.

Specifically, with the exact same 12GB (3 x 4GB) DDR3-1600 9-9-9-24 RAM, the system with the GT 240 had to be overclocked to 3.7GHz just to achieve the same amount of total time (297 seconds) as the system with the GTX 470 and an i7-950 CPU at its stock 3.2GHz (3.06 without Turbo) speed. Both of these systems were tested using a RAID 0 array with two modern 1TB 7200RPM hard drives. Remember the 317-second total time with the i7-920 system overclocked to 3.675GHz and equipped with the GT 240 and only 6GB of DDR3-1600 RAM (as listed in the PPBM5 results list on the PPBM5 web site under the name "Randall's Mod Rod")? In that particular system, doubling the amount of RAM reduced the total PPBM5 benchmark time by only 20 seconds. This confirmed my suspicion that Premiere Pro CS5 versions 5.0.2 and 5.0.3 are "friendlier" to those systems with less than 12GB of RAM than versions 5.0.0 and 5.0.1 were. And my experiment with my main i7-950 system with an additional 4GB of RAM -- 16GB versus 12GB -- revealed that the MPEG-2 DVD encoding time took less than half as long with 16GB as it did with 12GB.

Re: Best Graphics Card for CS5
 

Hi Randell,

Even though you are test with two identical systems, won't there be some minor timing differences?

What about when you do the test in the same computer, are you seeing a difference in rendering speed?

I ask, because I have run tests using two identical systems, same motherboard, harddrive, CPU, CPU speed, memory, memory speed, etc and I also saw a slight difference in rendering speed. For example, system 1 has the GT240 and system 2 had a GT460. System 2 rendered faster. This lead me to believe the GT460 was a little faster.

However, when I would test the 2 different video cards using the same computer, just swappng the cards out and reinstalling the NVidia driver, there was hardly any difference in the rendering speed. Maybe 3 to 4 seconds between the two on a 6 min timeline. This was much smaller that what I saw when using two identical systems. The difference I figured was due to Windows 7 processes.

I also ran a system timing test on both systems and it did report that while both systems has identical memory speeds, cpu speeds, etc. system 2 was reporting 4% faster on the timing tests.

I am interested in hearing more about your findings as I am ruinning on AMD systems and not Intel.

Best Regards
Dave Knarr

Re: Best Graphics Card for CS5
 

has anyone tired the GTX 570?

I'm in the middle of building my new system

i7 Sandy Bridge 2600k
asus maximus iv extreme
12GB of RAM
GTX 570
all water cooled
Blu ray Burner

all that stuff has wiped me out of money so ill be adding faster hard drives maybe RAID and anything else i can upgrade at a later date

Re: Best Graphics Card for CS5
 

I really want to upgrade my current system to run CS5.5 Production Premium, and want to include in this an upgrade of the video card.
Current System looks like this:

Windows 7 Professional 64 bit (just upgraded from Vista "Ultimate" 64 bit)
12GB DDR-3-1066Mhz SDRAM (6 DIMMs)
1GB ATI Radeon HD 4850 (2 DVI, HDMi, adpt.)
Intel i7-950 processor
Seagate 1.5TB 7200rpm SATA 3Gb/s C Drive for applications, etc.
Seagate 2.0TB 7200rpm SATA 3Gb/s K Drive for projects
WD 4TB 7200 rpm eSATA 3Gb/s F drive for back-up.

Can I upgrade to a catd like the Geforce GTX 570, or what would be my best bet?
Thanks in advance,
SW

Re: Best Graphics Card for CS5
 

Well your system is pretty good except video card. You could upgrade to GTX470 but it operates little too hot. 570 is much better. But I preferred Quadro4000 if you edit AVCHD that is the best choice. I know it is pricy but it is worth it.
If you chose to go with 570 or 4000, this is just a reminder, they are not officially on the adobe list for supported cards but you can add them very easily and enjoy the performance.

Re: Best Graphics Card for CS5
 

Latest list of supported cards shown towards the bottom in their tech specs page.

Tech specs | Adobe Premiere Pro CS5.5

4000 and 570 are supported.

Re: Best Graphics Card for CS5
 

How can I find out if the GTX570 is compatible with my other components? Or is that an issue?

Re: Best Graphics Card for CS5
 

You just need to make sure that your PC's power supply unit (PSU) can handle the load. Some cheap 750W and 850W PSUs can't really handle more than about 500W to 600W without going ape.

Re: Best Graphics Card for CS5
 

I'm having trouble analyzing all of these results. How is it that a really cheap card with not so many cores can perform almost as well as a card with nearly double the cores? That guy got an email from Adobe and they told him that the more cores you have, the better/faster Adobe stuff will perform. Well these results are showing that, but... not nearly as much. You'd think doubling your cores = doubling performance. Why is this not so?

I'm still unsure of which card to get. If for some reason Adobe steps it up and releases an update where it can utilize those other cores, then I'll get a 570 instead of a cheaper card. But if for some reason it won't matter, I'll just get a cheap 260 or something.

Re: Best Graphics Card for CS5
 

I have the GTX570, it works great!

Re: Best Graphics Card for CS5
 

Jared, my experience is that to saturate even the cheap hacked GTX460 I had to add about ten GPU supported effects to a single clip - then I got 100% GPU use, and Adobe was stopping playback after a few moments. I guess up until this point it was the CPU/HDD that was the bottleneck as far as performance is concerned.

Therefore what kind of card you need is determined by how many effects you actually use. GPU does not accelerate decoding and playback of a clip itself.

Re: Best Graphics Card for CS5
 

According to some non-scientific tests, it is estimated that only about 100 cores are actually used so even if a card has triple the cores, only 100 of them are being used. So performance may increase in the future when more cores are utilized but it doesn't make that big of a difference currently.

Re: Best Graphics Card for CS5
 

Thanks a lot for the replies, appreciated. I think I MIGHT still go with a 570 just to have it future-proofed a little. There are still a couple things I'm unclear on. On the Adobe Cuda page, it says that encoding and decoding are not gpu-accelerated. I should know this by now, but what is the difference between rendering, decoding, and encoding in CS5.5 and where exactly do all of these take place?

As far as I know with a Cuda card, the preview window is sped up, some FX are sped up when using in real time, and the final rendering output is sped up as well. But the encoding/decoding thing confuses me.

So when you say playback, you're talking about playing it back in the Preview window correct? I'm not sure of the differences between hitting Enter, 0 on the keypad, and spacebar...

And thanks again for the replies, almost got this understood 100%.