Classic Mac Pro - Multi Cores vs 2011 MBP?

[iOrbit]

hi guys, just a question about the reality of performance differences in the context of encoding blu rays with handbrake.

I have a late 2011 MacBook Pro, i7 2.2ghz (2675QM)

I want to buy an older 'Classic' Mac Pro, either 2010 or 2012 model with atleast 6 cores, maybe 8 or even 12.

my question is, specifically -

when it comes to encoding a blu ray with handbrake (which takes on average 4 hours with my MacBook) is there any performance difference using a Mac Pro from 2010 or 2012?

i understand that my MBP i7 may be faster at single core processing than the Mac Pro at single core processing, but when it comes to the handbrake encoding, will multiple cores encode significantly faster? what difference can i expect?

i'd really appreciate your input,

thanks in advance

 

[Raunien]

when it comes to encoding a blu ray with handbrake (which takes on average 4 hours with my MacBook) is there any performance difference using a Mac Pro from 2010 or 2012?

The 2010 and 2012 machines are essentially the same machines. One is newer than the other in terms of how long it's been out, but they're the same machine.

i understand that my MBP i7 may be faster at single core processing than the Mac Pro at single core processing, but when it comes to the handbrake encoding, will multiple cores encode significantly faster? what difference can i expect?

thanks in advance

When it comes to handbrake/video encoding, more cores is always better and faster. If you want to go as fast as possible, get an 8 core machine and then upgrade the to dual 3.46ghz 6 core cpu's for a total of 12 cores.

 

[h9826790]

It depends on your encoding parameter. Also, since you said that's for blu ray encoding, so most likely are high resolution, high bitrate, long movie encoding. In this case, Handbrake should able to fully utilise all cores, and it's performance is almost directly related to the CPU raw performance.

For 2675QM, compare to the X5690 or W3690. I will expect a 100% improvement. For a Dual X5690, I will expect a 300% improvement.

In fact, when I upgrade from my W3520 (Quad 2.66) to W3690 (Hex 3.46). I did see this significant performance increment in Handbrake (especailly on the 2nd phase of a 2 pass encoding).

Besides, in normal situation, even though you run handbrake 24/7, the cMP won't suffer from any thermal throttling. So, it's another benefit to run Handbrake on the cMP.

 

[iOrbit]

The 2010 and 2012 machines are essentially the same machines. One is newer than the other in terms of how long it's been out, but they're the same machine.

When it comes to handbrake/video encoding, more cores is always better and faster. If you want to go as fast as possible, get an 8 core machine and then upgrade the to dual 3.46ghz 6 core cpu's for a total of 12 cores.

The only reason i'd buy a 2012 mac pro is because my experience of building my own computers is limited, and i am intimidated at the thought of upgrading my own CPU's (the HDD/GPU/RAM/Expansion is not intimidating for me though). So if i can get a 2012 classic Mac Pro with 2012 spec'd CPU's it would be easier, but probably not cheaper.

How much would it cost to upgrade those CPUs?

It depends on your encoding parameter. Also, since you said that's for blu ray encoding, so most likely are high resolution, high bitrate, long movie encoding. In this case, Handbrake should able to fully utilise all cores, and it's performance is almost directly related to the CPU raw performance.

For 2675QM, compare to the X5690 or W3690. I will expect a 100% improvement. For a Dual X5690, I will expect a 300% improvement.

In fact, when I upgrade from my W3520 (Quad 2.66) to W3690 (Hex 3.46). I did see this significant performance increment in Handbrake (especailly on the 2nd phase of a 2 pass encoding).

Besides, in normal situation, even though you run handbrake 24/7, the cMP won't suffer from any thermal throttling. So, it's another benefit to run Handbrake on the cMP.

I do have a budget, so its not particularly likely i will get the best possible CPU's that these Mac Pro's can use, all i know is i won't go less than 1 6 core CPU or 2 CPUs (8 Core), 12 would be awesome but not likely for me to afford.

lets say i end up with a 2010 MP - 2x 2.4ghz quad-core "Gulftown" (E5620) CPU's, if it takes 4 hours on my MacBook Pro (i apologise for lacking details) what might i possibly expect duration wise from this Mac Pro?




thanks so much for your help guys, i'd still appreciate any additional input from others- experience, own comparisons etc. please feel welcome.

 

[Raunien]

The only reason i'd buy a 2012 mac pro is because my experience of building my own computers is limited, and i am intimidated at the thought of upgrading my own CPU's (the HDD/GPU/RAM/Expansion is not intimidating for me though). So if i can get a 2012 classic Mac Pro with 2012 spec'd CPU's it would be easier, but probably not cheaper.

How much would it cost to upgrade those CPUs?




I do have a budget, so its not particularly likely i will get the best possible CPU's that these Mac Pro's can use, all i know is i won't go less than 1 6 core CPU or 2 CPUs (8 Core), 12 would be awesome but not likely for me to afford.

lets say i end up with a 2010 MP - 2x 2.4ghz quad-core "Gulftown" (E5620) CPU's, if it takes 4 hours on my MacBook Pro (i apologise for lacking details) what might i possibly expect duration wise from this Mac Pro?




thanks so much for your help guys, i'd still appreciate any additional input from others- experience, own comparisons etc. please feel welcome.

2 x 3.46 6 core's (http://ark.intel.com/products/52576/Intel-Xeon-Processor-X5690-12M-Cache-3_46-GHz-6_40-GTs-Intel-QPI) would probably cost around $500-600.

In terms of how much faster, it's hard to say. Try finding someone with a similar config and having them run the conversion.

 

[h9826790]

lets say i end up with a 2010 MP - 2x 2.4ghz quad-core "Gulftown" (E5620) CPU's, if it takes 4 hours on my MacBook Pro (i apologise for lacking details) what might i possibly expect duration wise from this Mac Pro?

I am quite sure a 2010 dual 2.4 Quad is more expensive than the 2009 single Hex 3.46. But the W3690 should be tiny bit faster than the dual E5620. [For info, the 2010 (5,1) and 2009 (4,1) are technically the same machine, there is a firmware hack to make the 4,1 become a 5,1].

So, if your want the performance now but lower the cost as much as possible. Go for a single 2009, and then get a W3680 (Hex 3.33). It's not too hard to get this config by $600. [A single W3680 still faster than dual E5620]

And I will expect the cMP can finish the same job in around 2 hours. Again, this is really hard to tell. If you have a short movie for me to test, I can tell you my result. But you have to give me all the encoding parameter. A general handbrake performance test may not able to reflect your work flow.

However, once you go for the single CPU model, there is no low cost way that you can go for the dual CPU setup. You basically have to buy another dual CPU model for that.

 

[iOrbit]

2 x 3.46 6 core's (http://ark.intel.com/products/52576/Intel-Xeon-Processor-X5690-12M-Cache-3_46-GHz-6_40-GTs-Intel-QPI) would probably cost around $500-600.

In terms of how much faster, it's hard to say. Try finding someone with a similar config and having them run the conversion.

Thanks, Raunien.

And I will expect the cMP can finish the same job in around 2 hours. Again, this is really hard to tell. If you have a short movie for me to test, I can tell you my result. But you have to give me all the encoding parameter. A general handbrake performance test may not able to reflect your work flow.

However, once you go for the single CPU model, there is no low cost way that you can go for the dual CPU setup. You basically have to buy another dual CPU model for that.

Sorry its taken me awhile to get back to you,

i've been looking at prices on ebay and you're about right, the 6 core can go for a bit cheaper than the dual (8 core).

i did a little test today, and have grown a bit skeptical about performance upgrade, hope you can tell me what you think.

*edit* i have now completed a second test and my skepticism has been lifted!

Does geekbench scores not show a true reflection of encoding times when we're comparing physical cores? or does it make no difference how many physical cores there are for encoding if another CPU setup with less cores matches it for performance?

I was doing an encode test on a 1080p mkv blu ray rip (Generation Kill: Combat Jack, 01:07:00s) on a Late 2013, 27inch iMac @ 3.5ghz Quadcore i7 / 16GB 1600mhz DDR3 / 3TB Fusion Drive / 4GB GTX 780M (maxed i think)

HB Settings:

High Profile
h264
Constant Framerate (same as source)
preset; very slow
tune; film
profile; high
level;4.1
CRF; 20

audio;
DTS-HD MA 5.1:
1. AAC Stereo @ 160
2. AC3 5.1 @ 640
3. Auto Passthru

DTS 2.0
AC3 Stereo @ 224

and I'm getting about 3.5 - 4 hours on the HB estimation (haven't finished it to check how accurate that is)

i'm not sure what time it will be on my MacBook Pro, i will check it after this post and get back to you.

it made me wonder though, if my MacBook Pro isn't much slower, then can i really expect a cMP to be much better than this iMac when this iMac has this geek bench score;

single core; 3441
multi core; 12866


and cMP's similar to whats been discussed in the thread have;

Mid 2010 cMP - @3.33Ghz (6 core) W3680
sc; 2520
mc; 13953

Mid 2010 cMP - @ 2.40Ghz (8 core) E5620

sc; 1908
mc; 14159


my MacBook Pro;

sc; 2173
mc; 8301

So it turns out, on my MBP - i was getting a quote of 7.5 hours for the same job! quite impressive difference, i guess it just so happens that i ran a test on a very demanding film with more action/grain or something that is outputting a larger file and taking a longer time than the other tests i did a few weeks back on another film. I'll test one of the ones i did before on this iMac.. because if I'm going to get a Mac of this performance level, i want it to cut down my encode times from 4 5 hours right down to 2 or less.

also, is it inaccurate for me to compare a '3.5' ghz i7 cpu from this iMac to 2 x 2.40Ghz CPUs? will it give me basically the same encoding times so long as the geek bench scores are the same? (i mean this purely based on CPUs, i understand that the iMac and MBP have thermal throttling limitations compared to a Tower Mac Pro.) I'm just wondering in principle if a faster Processor (3.5ghz) will encode quicker with Handbrake than two 2.40ghz 8 cores total cpu set up, despite their geek bench scores being about the same?

@Raunien