Mavericks on a Mac Pro 2012

[stockholm80]

I installed OSX Mavericks on my Mac Pro (2012, 32Gb, 6-core 3.46GHz, ATI 5770) yesterday and I am not very happy with the experience so far. I'm on the latest developer preview DP3.

Stability seems to be good and most programs are working but the performance has taken a dive compared to Mountain Lion 10.8.4 on my system. My Geekbench result with Mountain Lion is 16200. With a clean install of Mavericks (from a thumb drive) with no third party applications installed I get 15600.

I thought it might be a compatibility problem with Geekbench so I tried opening a few of my projects in software such as FCPX and Logic Pro to test real-world performance. I found the difference to be very noticeable, both programs perform worse than in Mountain Lion in terms of cpu performance.

Anyone have an idea what is causing this? Is Mavericks optimized for newer technology than the 2012 Mac Pro (which is essentially a 2010). Has anyone else experienced the same thing?

 

[wallysb01]

Stability seems to be good and most programs are working but the performance has taken a significant dive compared to Mountain Lion 10.8.4 on my system. My Geekbench result with Mountain Lion is 16200. With a clean install of Mavericks (from a thumb drive) with no third party applications installed I get 15600.

Which of the tests are actually lower, because that's not very far off? It could just be system drive scores relating to the thumb drive.

I thought it might be a compatibility problem with Geekbench so I tried opening a few of my projects in software such as FCPX and Logic Pro to test real-world performance. I found the difference to be very noticeable, both programs perform much worse than in Mountain Lion in terms of cpu performance.

If you can spare the space, maybe you should make a partition on your normal boot drive and install it there and retest. I just don't think performance tests again a system bottom from a USB drive are going to be very accurate compared to a normal install.

 

[stockholm80]

Sorry to be unclear, I installed it clean from a thumb drive to my main disk which is a crucial M4 ssd. I was running it on the ssd when I did the benchmark.

In geekbench the processor integer performance was the one that showed the biggest difference compared to my mountain lion benchmark.

 

[mrapplegate]

In my opinion it's a little too early to be worrying so much about performance issues in a developer preview.

 

[Draeconis]

Sigh. 10.9 DP3 is just that, a Developer Preview. Don't expect it to be consistent/better/faster, it's a Developer Preview; it's for previewing the software, not for like-for-like production environment comparison.

10.9 GM probably won't be either. Given what a massive re-write this sounds like, don't be surprised about this either, who remembers what 10.6 was like? Total re-write = massive issues. By 10.6.2 everything was pretty much back to normal.

It's great that you have 10.9 DP3 give Apple as much feedback as possible, but posting here about performance issues tells everyone you don't understand what you're doing.

 

[stockholm80]

You are right that it shouldn't be compared to a final release. I'm just posting my observations. The reason was that I had seen so many posts here from people who are experiencing a great performance increase and I found it to be the opposite on my system.

 

[HenryDJP]

You are right that it shouldn't be compared to a final release. I'm just posting my observations. The reason was that I had seen so many posts here from people who are experiencing a great performance increase and I found it to be the opposite on my system.

With all due respect 16200 on ML to 15600 on Mavericks is hardly a difference for the way you're reacting about it.

 

[monkeybagel]

You are right that it shouldn't be compared to a final release. I'm just posting my observations. The reason was that I had seen so many posts here from people who are experiencing a great performance increase and I found it to be the opposite on my system.

Not seeing that on any of my machines; in fact, the opposite on a 2012 12-core Mac Pro 3.06/12GB/512GB SSD, 2012 MacBook Pro 2.7/8GB/650m/512GB SSD, and a 2013 MacBook Pro with Retina Display 2.7/16gb/650m/512GB SSD.

I am a registered developer, and have used this OS for a while as a production OS, without issue. Less problems than with 10.7 or 10.8 at this stage.

 

[stockholm80]

Not seeing that on any of my machines; in fact, the opposite on a 2012 12-core Mac Pro 3.06/12GB/512GB SSD, 2012 MacBook Pro 2.7/8GB/650m/512GB SSD, and a 2013 MacBook Pro with Retina Display 2.7/16gb/650m/512GB SSD.

Good to hear, maybe it's related to my specific system. Could you do a benchmark and verify? I notice an approx 4% drop in cpu performance on Geekbench which may seem small but when you are working with demanding software it is actually quite noticeable.

 

[stockholm80]

I just installed DP4 and my Geekbench scores are now up 600-700 points. Something must have changed in this update that has increased CPU performance on Mac Pro 6-cores. Any users that were experiencing a lower performance than in Mountain Lion on DP1-DP3 should install this update.

 

[3282868]

Not seeing that on any of my machines; in fact, the opposite on a 2012 12-core Mac Pro 3.06/12GB/512GB SSD

I have a sim. setup as you guys:

- 12-Core 5,1
- 32GB RAM
- 256GB OWC Mercury EXTREME Pro 6G (was hoping the next Pro would have SATA III )
- 4 SATA II HDDS (2 x 2TB, 3TB, 1TB)
- Radeon 5770
- 3 27" Cinema Displays

Preformed a clean install of 10.9, have booted into 10.8 and 10.6, tests don't seem much different. Many of 10.9's improvements are OpenGL Core 4.0+ support, efficiency (much like 10.6 was to 10.7), improved memory management, and new features here and there. Unfortunately, I wouldn't expect much in GeekBench scores. Someone made a valid point that currently GeekBench tests aren't very informative as system changes aren't necessarily in the categories GeekBench tests.

I have noticed a significant difference on my 2012 MacBook Air and 2012 MacMini, both with SSD's. The next Mac Pro seems to be more GPU dependent, seems Apple will be pushing workload onto the FirePro GPU which will be interesting in tests.

 

[stockholm80]

Unfortunately, I wouldn't expect much in GeekBench scores. Someone made a valid point that currently GeekBench tests aren't very informative as system changes aren't necessarily in the categories GeekBench tests.

Good to know. I do a test every time I upgrade my OS and in 10.6, 10.7 and 10.8 my Geekbench scores were very similar which confirms that.

In 10.9 DP3 the score dropped by 600 points so I was getting worried, but it seems that it was just developer preview related and is fixed now.

 

[sauria]

Update?

How is it running now, I have the same system and 10.9.1 seems very buggy.

I installed OSX Mavericks on my Mac Pro (2012, 32Gb, 6-core 3.46GHz, ATI 5770) yesterday and I am not very happy with the experience so far. I'm on the latest developer preview DP3.

Stability seems to be good and most programs are working but the performance has taken a dive compared to Mountain Lion 10.8.4 on my system. My Geekbench result with Mountain Lion is 16200. With a clean install of Mavericks (from a thumb drive) with no third party applications installed I get 15600.

I thought it might be a compatibility problem with Geekbench so I tried opening a few of my projects in software such as FCPX and Logic Pro to test real-world performance. I found the difference to be very noticeable, both programs perform worse than in Mountain Lion in terms of cpu performance.

Anyone have an idea what is causing this? Is Mavericks optimized for newer technology than the 2012 Mac Pro (which is essentially a 2010). Has anyone else experienced the same thing?