New mac pro purchase advice ASAP

[jb502]

Honestly, I just wanted to know which configuration would be faster, all around, than the 2.8 octo that I just sold.

I don't want to know which video card will give me more fps while playing WOW.

I use imovie, for quick videos, and premiere for more serious stuff. I use a panasonic tm700 that shoots 1080 30/60p. I do some simple edits, then export to either 720 or 1080p high quality preset.

Say a certain video took an hour to export on my 2.8 octo, what would it take to maybe cut it down to 45 minutes? or 50 minutes. or is that asking too much.

I just want an improvement.

I wish i could go someplace, have a 3.33 6 core, 2.4 8 core right next to each other, and see which one would export video faster with imovie and premiere.

 

[goMac]

I use imovie, for quick videos, and premiere for more serious stuff. I use a panasonic tm700 that shoots 1080 30/60p. I do some simple edits, then export to either 720 or 1080p high quality preset.

I would stick with Premiere if you're worried about encode times. iMovie's encoder is pretty awful.

Say a certain video took an hour to export on my 2.8 octo, what would it take to maybe cut it down to 45 minutes? or 50 minutes. or is that asking too much.

45 or 50 sounds fairly reasonable on the 2009. It's a reasonable mix of higher clockspeed and more cores. With turboboost on for clockspeed it would at least be the same clockspeed as the 2008, which should negate that difference.

 

[nanofrog]

I just sold my 2008 Mac Pro. It was 8 core 2.8ghz. I need to replace it. What I replace it with MUST be faster at EVERYTHING. 12 core is not an option. Too expensive. I am considering:

2010

3.33 6 core or 2.4 8 core

or

2009

2.66 octo

I can afford any of these, which should I get.

Mostly I need the speed for exporting from Imovie and Adobe Premiere. Wichever one will do these exports of 1080p fastest.

Are you taking upgrades into consideration in terms of your budget?

That is, you state a $3600 budget (from a later post), is that including additional RAM and disks?

Nehalems are quite a bit faster than Harpertowns. Disk speeds are pretty irrelevant as your CPU cannot encode as fast as your disk can write. Otherwise you could encode a full-size Blu-Ray in couple of minutes.

If the initial data is being read from an optical disk, that's the bottleneck (read speed of the optical disk is slower than an HDD write). The CPU or memory controller isn't an issue in such an instance.

How is that? Nehalems support 1066MHz DDR3 and triple-channeling while Harpertowns support only 800MHz DDR2 and dual-channeling.

Great.

But does the software make use of it (uses more memory bandwidth than what was available in the Harpertown)?

If it does, then the user will get an improvement in this area (bottlenecks in disk I/O will likely render it useless though, unless that's been addressed). Unfortunately, most software can't.

...so where exactly is the video he's encoding coming from then? Because if it's not a disk I'd like to know where.

Exactly.

Even if it's from a camera, it's still on some sort of storage media (i.e. DVD/BD, SD, CF, ...).

I'd argue that disk speeds matter for everything now, but that's a different debate.

I agree. There's precious little that doesn't rely on disk I/O.

You posted a benchmark for video cards. The results are in fps. This is a pointless benchmark for this topic

I have to agree with this for the most part, as the use of GPGPU processing isn't common yet, even though encoding is an area that could certainly benefit from it.

This will change in time, but it's not there just yet.

It's actually quite comparable. You're rendering video (i.e. you're rendering a certain number of frames per second), you're likely rendering effects from the GPU.... The big difference between video and games is that your disks enter into play.

Only if the encoding application utilizes GPGPU processing. Most still rely on CPU cores right now.

 

[goMac]

Only if the encoding application utilizes GPGPU processing. Most still rely on CPU cores right now.

Right, I think the breakdown is if anything needs to be done before the frame is encoded (i.e. rendering.)

 

[Evil Spoonman]

Depends. Are you encoding, or rendering AND encoding? If you're rendering, than a GPU benchmark is totally %100 applicable.

I assume you are talking about Premier here. In which case Mercury does indeed allow for GPU-acceleration of rendering. However, this is CUDA-based… and considering your only real options for that are the GTX 285 (NLA) and the Quattro cards (generally impractical for a hobbyist)… I would expect that to be a non-issue in this decision. Certainly linking to a Radeon 5870 and 5770 benchmark is of no help here.

Ehhhhh, H2.64 is still pretty seriously parallelizable, unless you've got like 3 keyframes.

Negative. H.264 encoding has a limited amount of parallelization. It doesn't even begin to touch the capability of my machine (500-600% per instance) when reading from an SSD. Keyframe density has some, but not a dramatic effect on this.

Performance per clock there is very little difference, almost all the differences are in the bus design.

Performance per clock arguments are somewhat difficult to make, and depend on the granularity with which you sample. In my case, a 2.8GHz Nehalem (H/T off) vs a 2.8GHz Harpertown - the Nehalem is appreciably faster at all operations. Better performance, same clock speed. If you go dig down and start dealing with IPC, things look different… Nehalem is able to decode more instructions per clock more of the time than Core. Technicalities aside we can agree that Nehalem is faster, isn't that what is important here?


OP: How many threads can Premier use to encode and/or render? Go ask the internet. If it is >8, the 2.26GHz Nehalem will begin to catch up. Mind you there are only 8 cores in that machine too, a virtual core is simply another pathway for data to an existing core. It is a net positive though, due to the width of the architecture's front end. What is the restriction in Premier's encoding process? Is it disk I/O, memory bandwidth? Answer these questions and you can more accurately target the hardware you need. Be it SSDs, memory, or processor architectures.

 

[nanofrog]

Right, I think the breakdown is if anything needs to be done before the frame is encoded (i.e. rendering.)

In the OP's case, I'm presuming the data is read directly off of the camera, which will be the bottleneck (rather than the raw footage imported to HDD/SSD, then edited and encoded). Either way, the camera is still slowing things down, and that cannot be changed. It would take a newer camera that has a faster I/O throughput to the MP to make a difference for that aspect.

But assuming the data is imported first (to HDD/SDD, then edited and encoded), the disk I/O bottleneck from this point on can be addressed (single disks for different tasks such as scratch, up to some sort of RAID configuration).

 

[jb502]

In the OP's case, I'm presuming the data is read directly off of the camera, which will be the bottleneck (rather than the raw footage imported to HDD/SSD, then edited and encoded). Either way, the camera is still slowing things down, and that cannot be changed. It would take a newer camera that has a faster I/O throughput to the MP to make a difference for that aspect.

But assuming the data is imported first (to HDD/SDD, then edited and encoded), the disk I/O bottleneck from this point on can be addressed (single disks for different tasks such as scratch, up to some sort of RAID configuration).

No. Video is first transferred to the hard drive

 

[nanofrog]

No. Video is first transferred to the hard drive

You can improve disk I/O, but without enough information to figure out your throughput needs (if you don't know directly), or budget, it's impossible to get you aimed in the right direction. Solutions could still be suggested, but they may not be that good a fit.

I'd suggest sitting down and figuring out the following (not necessarily all inclusive, but should be enough to get started):

1. How big are your files?
2. What is your current capacity requirement?
3. How much capacity do you consume per year?
4. What is your budget for upgrades?

Hobbyist v. professional use would be a good thing to know as well, to prevent spending more than is really needed.

Just something to consider before proceeding anyway...

 

[TheStrudel]

Investing in SSDs would probably have been a more dramatic speed improvement than any CPU increase will yield.

Unless you were already doing something similar to yoking all 8 cores to encode in compressor at the same time - which certainly does make things faster.

Sure, you can get a faster Mac Pro, but it seems to me like this is one of those cases where storage improvements will yield higher dividends.

 

[jb502]

Investing in SSDs would probably have been a more dramatic speed improvement than any CPU increase will yield.

Unless you were already doing something similar to yoking all 8 cores to encode in compressor at the same time - which certainly does make things faster.

Sure, you can get a faster Mac Pro, but it seems to me like this is one of those cases where storage improvements will yield higher dividends.

No matter which one I buy, it will be getting 12 gigs ram, and one or two ssd's.

I need some suggestions please, i am buying tommorrow.

please not more looking back at what I should have done to make the 2008 better. Its gone. I have to purchase something new tommorrow.

Hopefully someone will give me some advice that I asked for. Thanks.


Budget is anything except for the 12 cores.

 

[FluJunkie]

I'd like to say as buying advice threads go, this one has been a touch surreal.

Unfortunately OP, some of the best answers you could get rely on things like being able to see processor usage during your normal tasks - something you can't do with the sold Mac Pro.

I'm going to echo the suggestion that fast drives in a good configuration may be where you'll see the biggest speed boosts.

 

[gameface]

12GB of ram is way too little for either of your choices. You talking 1.5 - 2 GB per core. Hell, my laptop has 3 per core. Seriously, ram is your friend.

 

[jb502]

I'd like to say as buying advice threads go, this one has been a touch surreal.

Unfortunately OP, some of the best answers you could get rely on things like being able to see processor usage during your normal tasks - something you can't do with the sold Mac Pro.

I'm going to echo the suggestion that fast drives in a good configuration may be where you'll see the biggest speed boosts.

Yes. For the third time. I WILL be getting ssd's. My question is still, which configuration should I put them in ?

 

[nanofrog]

Yes. For the third time. I WILL be getting ssd's. My question is still, which configuration should I put them in ?

You're not giving sufficient information to really get you squared away properly (RAM capacity, how many disks, what kind, and how to configure them), but I'll put a potential solution out there...

SP Hex core system w/ 12GB of RAM (the RAM capacity is a stab in the dark, but is sufficient for a lot of users)

• SSD for OS/applications
• SSD for scratch/cache
• HDD for completed files
• HDD for a backup of completed files

Past this, more information is needed on your usage (size of the project files, ...).

 

[linuxcooldude]

I use imovie quite a lot on my 8 core mac pro. While im not nessicarily saying it might use all the cores in all cases, all the movies I've encoded uses all 8 cores. Well, might be other factors into play as well, i"m not sure.

 

[jb502]

can we just assume that disks will play no role in speed. Pretend my new setup will have the exact same disk setup as my 2008 octo.

If all other things are equal, what should I buy to be faster?

Only one person has made a suggestion, and that was the 3.33 6 core.

Does anybody else agree with this?

I will be purchasing in few hours.

 

[Evil Spoonman]

can we just assume that disks will play no role in speed. Pretend my new setup will have the exact same disk setup as my 2008 octo.

If all other things are equal, what should I buy to be faster?

Only one person has made a suggestion, and that was the 3.33 6 core.

Does anybody else agree with this?

I will be purchasing in few hours.

Completely agree that the 3.33GHz 6-core is the sweet spot. 12GB memory is also fairly on-target, but that can be taken care of after purchase.

 

[derbothaus]

God this thread is unnecessarily long. If you can't get the 12-core's then get the 6 core. The 8 is a dud and you have enough money to bypass the 4's.

 

[MattDSLR]

God this thread is unnecessarily long. If you can't get the 12-core's then get the 6 core. The 8 is a dud and you have enough money to bypass the 4's.

I don't know about that

it killed 6 core exporting 16 folders out of lightroom and resizing 40 gigs of files out of those folders at the same time
More core more power
Sorry for internet and other basic apps they both as fast
With new OS we will see more multicore advantages, plus easier upgrade in the future
8 core all the way

 

[allupons]

The 8 core is a machine without a purpose unless you are only buying it so that later you can spend thousands more to put 2 faster processors into the machine. For single/dual core apps, the 3.33 6 core is much faster than the 2.66 8 core. With fully multi-core optimized apps (and this more or less rarely/doesn't exist) the 3.33 6 core is virtually the same. More cores does not necessarily equal more raw power. The simple math (which is mostly a good representation of full multicore power) is 3.33 * 6 = 19.98 and 2.66 * 8 = 21.28. While 21.28 is slightly higher, bear in mind the it is using the somewhat dated nahelem architecture while the 6 core is using the newer westmere architecture, so it is basically a wash. The only thing the 8 core will definitely be faster with is addressing ram, as the the 2 processors it physically has provide more pipeline to the memory, but in the simplest terms possible the 6 core is faster or as fast than the 8 core at virtually everything. The 8 core is basically only really worth considering if you plan on replacing the 2.66 processors with much better, and that would cost you thousands.

 

[Hellhammer]

The 8 core is a machine without a purpose unless you are only buying it so that later you can spend thousands more to put 2 faster processors into the machine. For single/dual core apps, the 3.33 6 core is MUCH MUCH faster than the 2.4 8 core. With fully multi-core optimized apps (and this more or less rarely/doesn't exist) the 3.33 6 core IS STILL FASTER. More cores does not necessarily equal more raw power. The simple math (which is mostly a good representation of full multicore power) is 3.33 * 6 = 19.98 and 2.4 * 8 = 19.2. 19.98 > 19.2 for raw multithreaded power. The only thing the 8 core will be faster with is addressing ram, as the the 2 processors it physically has provide more pipeline to the memory, but in the simplest terms possible the 6 core is faster than the 8 core at nearly literally everything. There is a reason the 6 core costs more money than the 8 core after all. The 8 core is basically only worth considering if you plan on replacing the 2.4 processors with much better, and that would cost you thousands.

Maybe you should read the OP first as this is about 3.33GHz 6-core vs 2.66GHz 8-core. That's 21.28GHz using the same method you did.

 

[allupons]

Maybe you should read the OP first as this is about 3.33GHz 6-core vs 2.66GHz 8-core. That's 21.28GHz using the same method you did.

Ah, my bad for reading slightly too quickly and only noticing 6 core vs 8 core in responses at a glance. Given the slight boost from westmere architecture over the 2009 nahelem, you can replace the word faster for multithreaded apps with virtually the same speed. Therefore, for most every app that exists the 3.33 6 core will be much faster, and for a small handfull of apps they will be virtually the same speed. I will edit my original post to reflect this.

 

[jb502]

ok, well I did not purchase last night.

The next hexacore that pops up on the refurb store is getting purchased by me immediately.

I know they don't show up very often, but i am waiting until another one does since I can save about 600 bucks. I'v already ordered 4x4gb sticks of ram from owc

 

[allupons]

Also, to very explicitly answer your question about rendering HD video the fastest, barefeats.com has a table listing the render time for a 2min 30sec video using a multi-core 64bit rendering app. The results among the machines you are considering is 54 seconds for the 3.33 6 core, 66 seconds for the 2.4 8 core 2010 mac pro, and 72 seconds for the 2.66 8 core 2009 mac pro. So it seems westmere has a bigger raw advantage than I even knew as the 2.4 8 core was faster than the 2.66 8 core nahelem from 2009. Regardless, the 6 core was 122% faster than the 2.4 8 core and 133% faster than the 2.66 8 core for rendering video in the real world.

 

[jb502]

Also, to very explicitly answer your question about rendering HD video the fastest, barefeats.com has a table listing the render time for a 2min 30sec video using a multi-core 64bit rendering app. The results among the machines you are considering is 54 seconds for the 3.33 6 core, 66 seconds for the 2.4 8 core 2010 mac pro, and 72 seconds for the 2.66 8 core 2009 mac pro. So it seems westmere has a bigger raw advantage than I even knew as the 2.4 8 core was faster than the 2.66 8 core nahelem from 2009. Regardless, the 6 core was 122% faster than the 2.4 8 core and 133% faster than the 2.66 8 core for rendering video in the real world.

Link?

If this is correct, than the 2.4 8 core should also be faster than my old 2008 2.8 8 core. correct?