Welcome to the MacNN Forums.

If this is your first visit, be sure to check out the FAQ by clicking the link above. You may have to register before you can post: click the register link above to proceed. To start viewing messages, select the forum that you want to visit from the selection below.

You are here: MacNN Forums > Software - Troubleshooting and Discussion > macOS > Factors Why A Mac Doesn't Perform Tasks More Quickly?

Factors Why A Mac Doesn't Perform Tasks More Quickly?
Thread Tools
schalliol
Mac Elite
Join Date: Oct 1999
Location: Carmel, IN, USA
Status: Offline
Reply With Quote
Aug 18, 2008, 03:24 PM
 
A couple years ago I started a thread that is related to this one, but I couldn't reply to it because it was too old, so, as the system told me, I'm starting a new thread.

I have a reasonably powerful configuration at times with powerful CPUs, lots of RAM and a fast disks, and I find tasks can still take the machine quite a while with a CPU core not maxed to 100% utilization (let's not discuss multiple processing as a factor, since we all know that most apps can't use as many CPUs as you can give it), available ram and lots of available hard drive space. I know that some board components could be an issue, but what is an app doing if it's not finished performing all its functions while none of these components are fully utilized? Basically, what are the bottlenecks if not CPU, RAM, Hard Disks?
iMac Late '15 5K 27" 4.0 Quad i7 24/512GB SSD OWC ThunderDock 2 Blu-Ray ±RW MBP '14 Retina 15" 2.6 16/1TB iPhone 7+ 128 Jet Black iPad Pro 128 + Cellular

FOR SALE: MP '06 Yosemite 8x3.0 24/240GB SSD RAID 0, 240GB SSD, 1.5TB HDD RAID 0, 1TB HDD, Blu-Ray±RW, Radeon HD 5770
     
BLAZE_MkIV
Professional Poster
Join Date: Feb 2000
Location: Nashua NH, USA
Status: Offline
Reply With Quote
Aug 18, 2008, 03:29 PM
 
The time it takes to move data between the CPU and memory. Ran into this with a stl map and a pseudo random key (bad idea).
     
GSixZero
Mac Elite
Join Date: Aug 2004
Location: Seattle, WA
Status: Offline
Reply With Quote
Aug 18, 2008, 03:32 PM
 
Do you have examples of when this is happening?

When you're rendering a huge video, it could be bus or memory bottlenecks, when you're playing the latest FPS, it could be video card bottle necks, etc etc...

Even if you're doing pure CPU computation, often time one task relies on knowing what happened with previous tasks, limiting the efficiency of the CPU.

Really matters what you're doing to know where the bottleneck might be.

ImpulseResponse
     
Chuckit
Clinically Insane
Join Date: Oct 2001
Location: San Diego, CA, USA
Status: Offline
Reply With Quote
Aug 18, 2008, 03:39 PM
 
Could be network access. Could be the specific program you're using has a particular file or lock that it's contending for and it winds up spinning for a while, or a number of other software design decisions. What kind of tasks are you doing that seem to be taking a long time with no resource consumption?
Chuck
___
"Instead of either 'multi-talented' or 'multitalented' use 'bisexual'."
     
schalliol  (op)
Mac Elite
Join Date: Oct 1999
Location: Carmel, IN, USA
Status: Offline
Reply With Quote
Aug 18, 2008, 03:41 PM
 
There are a number of tasks where I've noticed this (across lots of machines) and I'll try to pick some up to show in the near-term in this thread. A good example that everyone faces is even application launching. Just launching Illustrator CS3 I see that during the launch, the total utilization is only 37% on average during the launch (and that's based on a single core). I know that the disk isn't being maxed, but perhaps there is a access time bottleneck there, and there's plenty of RAM. The only thing I can come up with is hard drive access time. I'll also see that at times with a huge sheet in Excel or a complex search in spotlight. This doesn't include the times where the machine or app is locked up. There are a number of "application" tasks that I'm just not thinking of, since, these are obviously more mundane, but they'll come back to me.
( Last edited by schalliol; Aug 18, 2008 at 03:50 PM. )
iMac Late '15 5K 27" 4.0 Quad i7 24/512GB SSD OWC ThunderDock 2 Blu-Ray ±RW MBP '14 Retina 15" 2.6 16/1TB iPhone 7+ 128 Jet Black iPad Pro 128 + Cellular

FOR SALE: MP '06 Yosemite 8x3.0 24/240GB SSD RAID 0, 240GB SSD, 1.5TB HDD RAID 0, 1TB HDD, Blu-Ray±RW, Radeon HD 5770
     
lenox
Senior User
Join Date: Aug 2003
Location: united states empire
Status: Offline
Reply With Quote
Aug 18, 2008, 04:32 PM
 
Yes, it does sound like hard disk bottlenecks are at fault here. You have a few options, including faster disks and dedicated RAID hardware. Prepare for some noise!
c2d 2.66ghz iMac
500gb/2gb/motu ultralite
     
BLAZE_MkIV
Professional Poster
Join Date: Feb 2000
Location: Nashua NH, USA
Status: Offline
Reply With Quote
Aug 18, 2008, 04:35 PM
 
Application launch is usually limited by disk access. Its spending 60% of its CPU time waiting for the disk and the disk is idle when the CPU is busy.
     
analogika
Posting Junkie
Join Date: Feb 2005
Location: 888500128
Status: Offline
Reply With Quote
Aug 18, 2008, 04:36 PM
 
download MenuMeters to get an idea of what exactly is going on.

Most times, it's memory swap that bogs things down.
     
lenox
Senior User
Join Date: Aug 2003
Location: united states empire
Status: Offline
Reply With Quote
Aug 18, 2008, 04:47 PM
 
Originally Posted by analogika View Post
Most times, it's memory swap that bogs things down.
If this is the case, then they should get more ram. Swap file should hopefully not be used at all, or very minimally.

Think about it: during the launch of a program, the data is loaded from hard drive to ram. If the ram is maxed out, it gets loaded ... back to hard drive, which acts as ram. Not a fun way to spend proc cycles, that's for sure
c2d 2.66ghz iMac
500gb/2gb/motu ultralite
     
analogika
Posting Junkie
Join Date: Feb 2005
Location: 888500128
Status: Offline
Reply With Quote
Aug 18, 2008, 05:08 PM
 
Originally Posted by lenox View Post
If this is the case, then they should get more ram.
I assure you, I'd love to.

(3 gig max)
     
mduell
Posting Junkie
Join Date: Oct 2005
Location: Houston, TX
Status: Offline
Reply With Quote
Aug 18, 2008, 07:21 PM
 
Latency, bandwidth, and Amdahl's law.
     
besson3c
Clinically Insane
Join Date: Mar 2001
Location: yes
Status: Offline
Reply With Quote
Aug 18, 2008, 07:53 PM
 
Why not run an iostat -w 1 during your next application launch to check for load average?
     
schalliol  (op)
Mac Elite
Join Date: Oct 1999
Location: Carmel, IN, USA
Status: Offline
Reply With Quote
Aug 19, 2008, 01:37 AM
 
Great responses, everyone! I'll check the terminal command on my launches and other cases where I find that that an app isn't working efficiently.
iMac Late '15 5K 27" 4.0 Quad i7 24/512GB SSD OWC ThunderDock 2 Blu-Ray ±RW MBP '14 Retina 15" 2.6 16/1TB iPhone 7+ 128 Jet Black iPad Pro 128 + Cellular

FOR SALE: MP '06 Yosemite 8x3.0 24/240GB SSD RAID 0, 240GB SSD, 1.5TB HDD RAID 0, 1TB HDD, Blu-Ray±RW, Radeon HD 5770
     
schalliol  (op)
Mac Elite
Join Date: Oct 1999
Location: Carmel, IN, USA
Status: Offline
Reply With Quote
Aug 21, 2008, 11:18 AM
 
Ok, here's a scenario. I had a big Excel spreadsheet. In it, I was copying 5 rows and inserting them in between a couple of other rows. It took 7-8 seconds to do this each time, and a number were inserted on the 8-core listed below (with Barracuda ES RAID 0 startup volume). I was only doing this and no other major processing at the time, so it ties up Excel for a while. iStatMenus showed available RAM and no core at 100%. Here's the output from the iostat -w 1 command:

Code:
disk0 disk1 disk2 cpu load average KB/t tps MB/s KB/t tps MB/s KB/t tps MB/s us sy id 1m 5m 15m 0.00 0 0.00 6.00 172 1.01 6.46 155 0.98 15 3 82 0.91 0.90 0.68 0.00 0 0.00 6.72 209 1.37 6.50 219 1.39 15 3 82 0.91 0.90 0.68 0.00 0 0.00 6.70 224 1.46 6.72 224 1.47 15 3 82 0.91 0.90 0.68 0.00 0 0.00 7.47 194 1.41 7.77 185 1.40 15 4 81 0.92 0.90 0.68 0.00 0 0.00 5.33 123 0.64 5.05 111 0.55 15 3 82 0.92 0.90 0.68 0.00 0 0.00 5.70 236 1.31 5.84 194 1.11 14 3 83 0.92 0.90 0.68 0.00 0 0.00 5.29 202 1.04 5.49 202 1.08 16 3 81 0.92 0.90 0.68 0.00 0 0.00 6.93 146 0.99 6.79 164 1.09 15 3 82 0.92 0.90 0.68 0.00 0 0.00 6.61 115 0.74 6.35 130 0.81 14 3 83 1.08 0.93 0.69 0.00 0 0.00 11.84 93 1.07 11.72 81 0.93 15 4 81 1.08 0.93 0.69 0.00 0 0.00 5.20 199 1.01 5.12 208 1.04 14 4 82 1.08 0.93 0.69 0.00 0 0.00 7.58 122 0.90 7.74 116 0.88 15 4 81 1.08 0.93 0.69 0.00 0 0.00 5.29 146 0.75 5.61 137 0.75 20 5 75 1.08 0.93 0.69 0.00 0 0.00 5.87 234 1.34 5.70 215 1.20 15 3 82 1.08 0.93 0.69 0.00 0 0.00 6.10 198 1.18 6.21 187 1.13 15 4 81 1.08 0.93 0.69 0.00 0 0.00 5.47 232 1.24 5.28 219 1.13 15 4 82 1.08 0.93 0.69
Also, while launching apps isn't really a problem or anything (and mainly I'm just curious) here's load for MS Word:
Code:
disk0 disk1 disk2 cpu load average KB/t tps MB/s KB/t tps MB/s KB/t tps MB/s us sy id 1m 5m 15m 8.00 9 0.07 5.16 124 0.62 5.10 106 0.53 10 7 83 0.67 0.68 0.66 14.06 69 0.95 9.28 264 2.39 8.89 288 2.50 9 8 83 0.67 0.68 0.66 8.00 16 0.12 12.74 125 1.55 13.59 113 1.50 16 7 77 0.67 0.68 0.66 13.73 73 0.98 13.21 134 1.73 12.71 143 1.77 10 5 84 0.69 0.69 0.66 8.00 13 0.10 15.45 137 2.06 16.00 132 2.06 6 6 89 0.69 0.69 0.66 13.73 73 0.98 10.93 209 2.23 9.69 254 2.40 3 5 92 0.69 0.69 0.66 13.89 71 0.96 16.93 100 1.65 14.65 127 1.82 3 5 92 0.69 0.69 0.66 8.00 9 0.07 13.71 178 2.38 14.38 160 2.24 4 4 93 0.69 0.69 0.66 13.50 76 1.00 8.89 199 1.73 8.67 199 1.68 4 5 91 0.80 0.71 0.67 8.00 17 0.13 10.42 144 1.46 10.98 141 1.51 9 6 86 0.80 0.71 0.67 13.36 78 1.02 8.85 199 1.72 8.96 205 1.79 5 6 89 0.80 0.71 0.67 8.00 13 0.10 5.77 88 0.50 5.66 82 0.45 4 4 92 0.80 0.71 0.67 14.33 66 0.92 11.64 218 2.48 11.32 217 2.40 3 4 93 0.80 0.71 0.67 8.00 14 0.11 8.52 158 1.31 7.79 185 1.41 3 5 93 0.73 0.70 0.66 13.65 74 0.99 12.23 105 1.25 11.28 132 1.45 9 6 85 0.73 0.70 0.66
And DreamWeaver CS4ß, which took about 25 seconds to launch:
Code:
disk0 disk1 disk2 cpu load average KB/t tps MB/s KB/t tps MB/s KB/t tps MB/s us sy id 1m 5m 15m 7.33 6 0.04 13.41 182 2.38 14.99 172 2.51 5 3 92 0.43 0.59 0.62 4.80 10 0.05 15.60 190 2.89 17.25 173 2.91 4 3 93 0.43 0.59 0.62 4.00 7 0.03 42.05 139 5.70 38.64 155 5.84 3 3 93 0.43 0.59 0.62 5.33 9 0.05 15.78 232 3.57 16.35 220 3.51 4 3 93 0.43 0.59 0.62 5.60 10 0.05 15.88 163 2.53 15.22 169 2.51 7 5 88 0.43 0.59 0.62 6.00 6 0.04 22.93 138 3.09 21.12 151 3.11 3 3 94 0.39 0.58 0.62 4.44 9 0.04 20.46 207 4.13 22.63 180 3.98 4 3 93 0.39 0.58 0.62 4.80 10 0.05 18.49 175 3.16 16.86 196 3.22 2 2 96 0.39 0.58 0.62 4.50 8 0.04 13.84 164 2.21 15.54 138 2.09 4 3 93 0.39 0.58 0.62 5.33 9 0.05 11.13 191 2.07 11.35 193 2.14 5 3 92 0.39 0.58 0.62 5.00 8 0.04 9.15 233 2.08 9.34 222 2.02 3 3 94 0.36 0.57 0.61 4.00 9 0.04 10.81 177 1.87 10.64 179 1.86 3 3 93 0.36 0.57 0.61 4.50 8 0.04 11.66 138 1.57 10.91 164 1.75 5 6 90 0.36 0.57 0.61 5.14 14 0.07 8.23 174 1.40 8.43 167 1.37 6 11 83 0.36 0.57 0.61 5.60 5 0.03 12.66 128 1.58 12.63 127 1.57 5 5 90 0.36 0.57 0.61 4.50 8 0.04 6.94 217 1.47 6.71 222 1.45 4 3 93 0.41 0.57 0.61 4.80 10 0.05 9.09 173 1.53 8.49 185 1.53 3 3 93 0.41 0.57 0.61 5.45 11 0.06 11.83 183 2.11 11.96 176 2.05 5 4 91 0.41 0.57 0.61 4.67 6 0.03 20.65 131 2.64 20.21 135 2.66 3 3 94 0.41 0.57 0.61 5.14 7 0.04 16.01 176 2.75 14.78 194 2.80 9 5 86 0.41 0.57 0.61 5.00 8 0.04 6.74 174 1.14 6.60 179 1.15 5 4 91 0.46 0.58 0.62 4.00 5 0.02 7.86 218 1.67 8.10 208 1.64 3 3 94 0.46 0.58 0.62 15.68 68 1.04 8.24 124 1.00 7.93 137 1.06 3 3 94 0.46 0.58 0.62 4.63 19 0.09 6.21 234 1.42 6.25 216 1.32 4 4 92 0.46 0.58 0.62 5.11 18 0.09 6.34 137 0.85 6.40 129 0.81 4 3 93 0.46 0.58 0.62 5.28 25 0.13 5.04 243 1.19 5.38 192 1.01 3 4 93 0.42 0.57 0.61 6.00 4 0.02 10.43 69 0.70 9.95 74 0.72 2 2 95 0.42 0.57 0.61 4.89 9 0.04 8.39 156 1.28 8.20 167 1.34 6 4 90 0.42 0.57 0.61 4.00 7 0.03 6.12 266 1.59 6.07 278 1.65 3 3 94 0.42 0.57 0.61 5.00 8 0.04 17.59 148 2.54 22.43 109 2.39 5 3 92 0.42 0.57 0.61 4.57 7 0.03 5.25 299 1.53 5.13 277 1.39 3 3 94 0.39 0.56 0.61 7.00 8 0.05 6.59 167 1.07 6.48 147 0.93 6 3 90 0.39 0.56 0.61
With an untrained eye, this looks like minimal data to move. If the buses aren't saturated, the memory is available, the CPU is available, little data needs to be pulled from the HD, what would cause these to not execute more quickly? I don't have a good understanding of how to measure that the system could be waiting to find locations on the hard disks, but that wouldn't show up as throughput, which seems like a possible culprit. I'm not suggesting that these operations are performing slowly, but rather interested in why they aren't faster. Any ideas?
( Last edited by schalliol; Aug 21, 2008 at 11:26 AM. Reason: Added Word and Dreamweaver Launch Info)
iMac Late '15 5K 27" 4.0 Quad i7 24/512GB SSD OWC ThunderDock 2 Blu-Ray ±RW MBP '14 Retina 15" 2.6 16/1TB iPhone 7+ 128 Jet Black iPad Pro 128 + Cellular

FOR SALE: MP '06 Yosemite 8x3.0 24/240GB SSD RAID 0, 240GB SSD, 1.5TB HDD RAID 0, 1TB HDD, Blu-Ray±RW, Radeon HD 5770
     
slpdLoad
Mac Elite
Join Date: Jun 2006
Status: Offline
Reply With Quote
Aug 21, 2008, 11:37 AM
 
I don't have a whole lot to add here, except to commiserate with you. Technical excuses aside, it's certainly frustrating to have an immense amount of horsepower sitting at your desk, but you hardly ever get to use more than 10% of it. It feels like such a waste that I should ever have to wait for any type of mundane task, when that much of my system appears to be unable to work just that much harder to get stuff done.

Edit: Now that I think about it, isn't this how scientists say our brains work too? We really are only using a small fraction of their true capacity most of the time.
     
schalliol  (op)
Mac Elite
Join Date: Oct 1999
Location: Carmel, IN, USA
Status: Offline
Reply With Quote
Aug 21, 2008, 12:00 PM
 
Originally Posted by slpdLoad View Post
I don't have a whole lot to add here, except to commiserate with you. Technical excuses aside, it's certainly frustrating to have an immense amount of horsepower sitting at your desk, but you hardly ever get to use more than 10% of it. It feels like such a waste that I should ever have to wait for any type of mundane task, when that much of my system appears to be unable to work just that much harder to get stuff done.

Edit: Now that I think about it, isn't this how scientists say our brains work too? We really are only using a small fraction of their true capacity most of the time.
Yeah, it's a bummer indeed, however I don't mean to suggest that the machine is sitting at 10% all the time. With good MP aware apps or a number of simultaneous activities the system is pretty well utilized and performs well. That said, waiting for these mundane activities is a bummer when it seems that those should be faster.
iMac Late '15 5K 27" 4.0 Quad i7 24/512GB SSD OWC ThunderDock 2 Blu-Ray ±RW MBP '14 Retina 15" 2.6 16/1TB iPhone 7+ 128 Jet Black iPad Pro 128 + Cellular

FOR SALE: MP '06 Yosemite 8x3.0 24/240GB SSD RAID 0, 240GB SSD, 1.5TB HDD RAID 0, 1TB HDD, Blu-Ray±RW, Radeon HD 5770
     
lenox
Senior User
Join Date: Aug 2003
Location: united states empire
Status: Offline
Reply With Quote
Aug 21, 2008, 12:07 PM
 
Looks like earlier responses still apply. Perhaps startup times could be sped up if the makers of the software you use wrote multithreaded startup routines...though I am not sure if the copying/pasting in Excel could really be 'optimized' for that sort of situation.

One should keep in mind that not all code is meant to utilize all available resources 100% of the time. Usually, that's actually a sign that something is not playing nice, as it leaves other processes starved for resources and makes the system feel unresponsive. Is it possible that the situations you describe could get done faster without rendering the system unresponsive, if the developers 'optimized' their code? Sure...probably. But - you said it yourself. These launch times aren't really a problem. There's no drive to make them faster as they don't really take too long and, on the flip-side, they don't render your system unresponsive, causing your network connections to drop and whatnot.
c2d 2.66ghz iMac
500gb/2gb/motu ultralite
     
TETENAL
Addicted to MacNN
Join Date: Aug 2004
Location: FFM
Status: Offline
Reply With Quote
Aug 21, 2008, 12:09 PM
 
Originally Posted by slpdLoad View Post
Now that I think about it, isn't this how scientists say our brains work too? We really are only using a small fraction of their true capacity most of the time.
That's what scientologists say – not scientists. A small but decisive difference.
     
Catfish_Man
Mac Elite
Join Date: Aug 2001
Status: Offline
Reply With Quote
Aug 23, 2008, 04:41 PM
 
The best way to determine why something is going slow is Shark, which is included with Apple's CHUD tools. Interpreting the output can be a little tricky though.
     
patrix
Junior Member
Join Date: Sep 2006
Status: Offline
Reply With Quote
Aug 23, 2008, 05:13 PM
 
Originally Posted by TETENAL View Post
That's what scientologists say – not scientists. A small but decisive difference.
You know, I went on a google hunt to disprove you and find the scientific reports about that -- and I'm learning a lot of stuff now

Although it would still be true in the sense that we don't use 100% capacity of our brain all the time, so in the context of this discussion the analogy holds - not all the CPU's power is constantly being used

Still, lots of information out there, gonna learn a lot in the next few days!

Patrix.
     
   
Thread Tools
 
Forum Links
Forum Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts
BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Top
Privacy Policy
All times are GMT -4. The time now is 11:32 AM.
All contents of these forums © 1995-2017 MacNN. All rights reserved.
Branding + Design: www.gesamtbild.com
vBulletin v.3.8.8 © 2000-2017, Jelsoft Enterprises Ltd.,