Thursday, November 26, 2009

Kramer VS-30FW FireWire 800 Hub | Seagate FreeAgent Desk for Mac | SanDisk Extreme FireWire CompactFlash Reader | Transcend 600X CompactFlash

Because current iMacs only come with a single FireWire 800 port, a hub is often necessary if you have more than one FireWire device. Unfortunately, as I mentioned before, FW800 hubs are hard to come by. Luckily I was able to find the Kramer VS-30FW 3-port FireWire 800 hub for under US$80 at a reputable dealer, so I immediately bought one. I had been told that this hub is popular in professional audio/video environments. In fact, the VS-30FW can be rackmounted in batches of three according to the Kramer website, using their rack adapter.

The hub is a small metal device and seems solidly built. One thing I liked about the power adapter is that it has a screw collar on the plug, so that the plug cannot be pulled out by accident. That's a nice touch, that I've only seen before on higher end equipment. It only has three ports, which means it basically functions as an active 1 --> 2 splitter. One port is used for the input, and two ports are outputs.

I now have it hooked up to my quad-core iMac, splitting my single FW800 port to one FW800 chain and one FW400 chain. This is done with a Seagate FreeAgent Desk for Mac external FW800 drive, and a Belkin FireWire 400 hub hooked up to a Seagate FreeAgent Pro 750 GB external FW400 drive. Both drives were reasonably quiet. Sitting behind the computer, I could barely hear them, as they have no fans. I did notice a slight hum from vibrations transmitted through the table, but this was gone if I put the drives on top of a cloth.

The FreeAgent Desk for Mac is a multi interface drive, and I was thus able to test FW800, FW400 (through the Belkin hub), and USB 2.

The test consisted of a 3734.5 MB file transferred from the external drive to the iMac's internal 2 TB Hitachi HDS722020ALA330 SATA drive.

USB 2.0: 104.3 s or 35.8 MB/s (34.1 MiB/s), 286 Mbps
FW400: 93.5 s or 39.9 MB/s (38.1 MiB/s), 320 Mbps
FW800: 43.4 s or 86.0 MB/s (82.1 MiB/s), 688 Mbps

FW800 likely gets reasonably close to the drive's max transfer rates, while FW400 is less than half as fast, and USB 2 is slower still.

FW800 also can mean significantly greater speeds for CompactFlash downloads too. Using a SanDisk Extreme FireWire CompactFlash Reader (daisychained through the Seagate FW800 drive) with a 16 GB Transcend 600X UDMA CompactFlash card containing 15726.71 MB worth of Canon EOS 7D files, the speed difference was very obvious.

FW400: 405.2 s or 38.8 MB/s (37.0 MiB/s), 310 Mbps
FW800: 226.4 s or 69.5 MB/s (66.2 MiB/s), 556 Mbps

It would be nice Apple offered eSATA support for hard drives, but having FW800 is almost as good for non-RAIDed hard drives, and FW800 is very flexible for other usage as well like audio/video devices and CompactFlash readers too. One nice thing about FireWire is having that other FW400 drive hooked up didn't slow down the FW800 transfer speeds. FW800 hardware can run at full speed in mixed FW400/FW800 setups. (See below for one exception.)

However, there were two problems I noticed during testing with the Seagate FreeAgent Desk for Mac drive.

1) If I turned on the option in the OS X System Preferences to allow the computer to put hard drives to sleep, sometimes the computer would hang if the Seagate drive were plugged in. Unchecking that option solved that problem.

2) Once after some testing, the transfer rate over FW800 from the Seagate drive dropped to just under 20 MB/s. Going back to FW400 gave me almost 40 MB/s. I couldn't figure out why, so I rebooted, and was able to get my 80+ MB/s speeds back. I don't know why this happened, but it has not happened again since.

Now I just await Apple's fix to the disappearing external FireWire drives caused by 10.6 Snow Leopard. My older external region free bit-set SuperDrive is not detected in Snow Leopard, but works fine in 10.5.8. I can get around this problem by taking the IOFirewireSerialBusProtocolTransport.kext from 10.5.8 and installing it in 10.6, but obviously, this is a hack. Apple is aware of this problem (Bug ID# 7201113) and is working on it.

I also wonder how long it will take Apple to implement USB 3.0, given that USB 3.0 peripherals are now out. My guess is 2011 - 2012, which may be about the time I'll want to replace my Core i7 iMac anyway.

Wednesday, November 25, 2009

iMac Core i7 power utilization

Using my trusty Kill-A-Watt power meter, these are the measurements I got for the iMac:

Off - 2 Watts
Sleep - 2-4 Watts
Idle, screen off - 55 Watts
Idle, min brightness - 71 Watts
Idle, mid brightness - 96 Watts
Idle, max brightness - 157 Watts

With screen set to mid brightness:

Booting up - Up to 184 Watts
Transfer file to iMac over Ethernet - 125 Watts
Cinebench OpenGL - Up to 142 Watts
Cinebench multi-CPU - Up to 196 Watts

With screen set to max brightness, the Cinebench multi-CPU benchmark can bring the iMac's total power utilization to 257 Watts.

This means the screen power can vary by almost 90 Watts between the lowest and highest brightness settings, and for the CPU, going from idle to near full usage, total power utilization will increase by 100 Watts. TDP of the Lynnfield Core i7-860 CPU is only 95 Watts (not 100) at max usage, but that doesn't include the increased power needs of other parts of the computer such as memory, etc. when the computer is run at full tilt. The power utilization would likely increase even more with heavier usage of the hard drive and GPU.

Given these numbers, we shouldn't feel the need to completely power off our Core i5/i7 iMacs to save energy. Sleep is good enough, as the difference in power usage between sleep and off is negligible. However, if you leave your computer on 24/7 without sleep, the machine will still use about 55 Watts (when the screen is off). At say 14 hours a day in this mode, that's 0.77 kilowatt-hours. At 11¢ per kilowatt-hour for example, that works out to about 8.5¢ per day, or $30 per year.

Handbrake 0.9.4 64-bit - 15% faster.

Handbrake 0.9.4 is out, and it finally comes in a 64-bit flavour. The Handbrake site states it's about 10% faster than the 32-bit version, but with my Spirited Away clip from my Core i7 iMac review, I'm getting a 15% boost in speed compared to 0.9.4 32-bit. These were my speeds with Handbrake 0.9.3.

The default setting on 0.9.4 is no longer the same, and neither are the settings in the Advanced section. This may be important since if you just change the main settings, 32-bit 0.9.4 seems much faster than 32-bit 0.9.3. However, if you take the default Advanced settings from 0.9.4 and use them with 0.9.3 as well, the speeds get much closer. I changed the main 0.9.4 settings to replicate the old 0.9.3 default - 1500 Kbps Average, 2-pass with Turbo first pass - and then re-benchmarked, but using the settings in the Advanced section from 0.9.4 for all tests.


This sped up encoding in 0.9.3 slightly, but the biggest gain was from using the 64-bit of Handbrake 0.9.4.

Handbrake 0.9.3 32-bit: 120.4 s (vs. 126.7 s with the 0.9.3 default)
Handbrake 0.9.4 32-bit: 121.6 s (1% slower)
Handbrake 0.9.4 64-bit: 105.5 s (14% faster)

64-bit Handbrake 0.9.4 is 15% faster than 32-bit Handbrake 0.9.4, and 14% faster than Handbrake 0.9.3. To put it in simpler terms, using these settings on similar animated material, it would only take 16 minutes to encode an entire 90 minute movie on my 2.8 GHz Core i7 iMac. Nice.

Well, sorta. If you adjust Handbrake 0.9.4's Advanced settings to the same flags that are default with 0.9.3 (ref=2:bframes=2:me=umh), then 0.9.4 slows right down. 32-bit 0.9.4 takes 196 seconds (or well over 3 minutes) to encode the same clip. Obviously the behaviour of the two versions with certain flags is quite different.

I should point out that Handbrake doesn't fully utilize all 8 virtual cores. However, it sometimes does come close.

Not bad for the first release of the 64-bit version of Handbrake.

Monday, November 23, 2009

Core i7 iMac - Benchmarks and first thoughts

I finally received my 27" Core i7 iMac last week, right before a killer busy weekend, so I haven't had as much time with it as I would have hoped. Nevertheless I've been able to run some comparative benchmarks and get some initial impressions of it.

Ergonomics 1: The height of the 27" iMac is the same as the old 24" iMac, and like the 24" iMac, the height is not adjustable. Unfortunately, this means that the ergonomics is still potentially problematic for some people. The height forces me to look upwards to see the top of the screen, as I am not tall. To put that in perspective, I have an external 24" screen that is height adjustable. Even if I raise that screen to its maximum height, which is too high for me, it's still shorter than the 27" and 24" iMacs. Raising the chair doesn't help, because it means taking my feet off the floor. Lowering the table can help in certain situations, that is if the table supports lowering. Either way, those represent ergonomic compromise. Fortunately for me, I don't need to sit at that computer 8 hours a day. If you're a person who is under 6 feet tall and need to use this computer all day long, you might want to consider a different setup. One such setup could be a 21.5" iMac and a second screen. The main problem with this is the lack of a Core i5/i7 option with the 21.5" models, and the lower end GPU. The other option is just to get a VESA adapter and mount to lower the 27" screen, but that is a fairly large chunk of change to spend just to get the screen a couple of inches lower.

Ergonomics 2: The pixel size really is quite small for a desktop computer. This wouldn't be so bad if Snow Leopard were resolution independent, but it isn't. When surfing with Safari, I found that initially my tendency was to lean forward to read the small text. Since then I have taken to increasing the zoomed size whenever I start Safari. The good news is that this zoom setting is retained for all sites. The bad news is that the zoom adjustment is very coarse grained, zoom slightly blurs images, and the zoom setting is reset when Safari quits or opens a new browser window. Firefox does remember zoom settings after a quit, and has more fine grained zoom control, but unfortunately (or fortunately depending on your perspective), the zoom setting varies from website to website. Each site has to be set independently. Again, here the 21.5" iMac has the advantage. The pixels are still more densely packed than the 24", but it's less severe than the 27". At least in the store, surfing Safari on the 21.5" at the native zoom setting was noticeably more comfortable than on the 27". When Apple adds a half decent GPU and quad-core to the 21.5", I might consider one to replace my 27".

Ergonomics 3: The Magic Mouse is an interesting design. Initially I did not like it that much but it is growing on me. The scroll function works very well, and is much nicer than the scroll ball on the Mighty Mouse. The scrolling feel is very much like the popular iPhone and iPod touch. Right clicking also works well. On the negative side, the mouse feels a bit narrow even for my relatively small hands, and the low profile and sharp edges can be uncomfortable. The lack of Exposé control is also a disappointment. Overall though, it seems to be a significant improvement over the Mighty Mouse. Let's just hope that its being Bluetooth doesn't end up be a problem like it was with my 24" iMac.

For the keyboard, I recommend getting the wired keyboard with numeric keypad, for the same price. It can be very annoying doing real work without a proper numeric keypad. The aluminum laptop-like keyboard isn't bad in terms of feel, but some of us still long for the old style IBM keyboards.

Screen quality: I have not yet formally calibrated the screen yet but I have been impressed with the quality of this IPS screen out of the box. It is certainly nicer than the Dell 2407WFP beside it, which after some attempted calibration still doesn't look as nice. The glossy screen is potentially a problem for some people, but luckily I prefer to work with my ambient room lights dimmed, so I don't notice many reflections.

The LED backlight is much more even than my previous non-LED 24", which suffered from mild corner dimness.

I have not encountered the much reported screen flicker issue. Some have attributed this to graphics drivers, but I have not seen it.

Speed: The machine feels noticeably faster than my previous Core 2 Duo model. Part of this may be due to the hard drive speed, but much of it seems to be the available memory and multiple fast cores. With (inexpensive) 8 GB RAM, the iMac almost never needs to page to disk with my usage, even if I have a virtual Windows XP machine and several OS X applications running in the background. Furthermore, with the 4 fast CPU cores, even just basic OS navigation is noticeably more responsive, as are most apps. Whereas the Core 2 Duo iMac was fast, the Core i7 is near instantaneous for many actions.

Compute intensive applications are hugely improved now. The extra efficiency per core and extra cores are very welcome with video encoding for example. My Core i7 is more than three times as fast as my older Core 2 Duo 2.33 GHz, for Handbrake H.264 encodes.

In this test I transcoded Chapter 7 of Spirited Away to H.264, from the computer's hard drive, and measured the total time to completion. At first I thought there might be something wrong with Handbrake 0.9.3 on Core i7, as I was getting speeds slower than my Core 2 Duo MacBook Pro. Then I realized it was because I had mistakenly installed the PowerPC version of Handbrake on my iMac. Yet even running under Rosetta, the speed of Handbrake was not terrible. The Intel version was however nearly 5X as fast, with the iMac 3.5X as fast as my MacBook Pro. It took just 2 minutes and 7 seconds to encode that 10 minute chapter using the default 2-pass x264 encoding settings. In other words, the iMac Core i7 can encode at nearly real-time speed under Rosetta, and nearly 5X real-time using native (32-bit) code.

The 3D types are going to be happy with Core i7 as well. Cinebench screams on this iMac.

The Core i7 iMac is 3.2X as fast as my MacBook Pro in Cinebench, and a mind-blowing 25X as fast as my iBook G4. Isn't 800% processor utilization on 8 (virtual) cores just sheer geek beauty?

Click picture to enlarge.

With all this speed, the current Mac Pro seems even more a ripoff than it did before. Core i7 Lynnfield in the iMac can usually keep up with the Bloomfield quad Mac Pro, for much, much less money. Unless there is an absolute need for full internal expandability right at this moment, then potential Mac Pro customers should wait until 2010 Q1, when 6-core and 12-core Gulftown based Mac Pros should make an appearance. Only at that time will the Mac Pro truly be worth a premium over the Core i7 iMac.

All in all, Apple has a big winner on its hands with the new Core i5 and Core i7 iMacs. However, a note to Apple: We've achieved enough already with iMac thinness. The next thing to conquer is iMac ergonomics, with either some height adjustability, or just lowering of the iMac's height in a fixed design, by decreasing the iMac's chin size. Oh and please add an eSATA port or two, too.