|
|
|
|
| FastMac Sonnet OWC CPUs |
| Written by Chris Tangora | ||||||||||
| Friday, 15 December 2006 | ||||||||||
|
Who's the fastest?
When I first took the hardware columnist position I wrote to several companies asking for products to review. One company that responded immediately was FastMac. They came through with a 1.47 GHz ($269.95) upgrade for my Digital Audio and a G4 upgrade to the CRT iMac next to me (review coming soon). I was set to do a review only on the FastMac upgrade when Vintage Computer of California sent over a Sonnet 1.33 GHz ($249) upgrade card for the Liquid Cooling System. Now we have two great Mac modding companies' offerings for getting your Mac up to speed. But wait, what's that? Other World Computing has a new offering, a sup'd up Newer Technology 2.0 GHz CPU ($399) that just landed in my lap. Let's take a look and see what kind of performance we get for the buck. 
The test system is a 733 MHz Digital Audio G4 with 1 GB RAM (2 x 512). All three came with their heatsink and fan already attached. Since my Mac was already half way disassembled I had an easy time installing them, but if you are starting out on a virgin system, it's still easy to get your system upgraded. All CPU upgrade cards work on a basic philosophy, out with the old, in with the new and they all come with detailed installation instructions. The one requirement for all upgrades is to update the firmware to 4.28, so you'll have to boot back into OS 9 and bring your Mac up to date before you can install. 
Sonnet 1.33 GHz Sonnet has a solid line for the Mac platform and is the most affordable CPU tested. With the 1.33 GHz card you are paying approximately $173 per Gig--not a bad price. This would be an exceptional improvement for the pre-500 MHz G4; the performance increase is noticeable and appreciated. Simple tasks that you grow accustomed to taking a while all of a sudden are shortened. 
Once installed the CPU worked as promised. It dramatically reduced the time needed for many tasks. I saw an immediate improvement in the performance and speed of applications. I started running some test across the CPU and it handled itself well with an average increase of 181% in processing power. The most noticeable change was the frames per second increase in iTunes visualizer (from 25 to 34 frames / second) and in the xBench scores for OpenGL and user interface. This upgrade made Aqua feel like it should with smooth transitions and non-skipping movies. The user interface test came back with a whopping 230 refreshes per second, the highest of all tested CPU's. An interesting result considering it is the "slowest" of all the tested processors. While there is no real logical explanation for this, I am assuming it is because of Sonnet's specific firmware update. For being the most economical, this processor definitely packed it's punch. 
However, despite this great boost, the installation of the Sonnet card was easily the most time consuming and frustrating. The screw holes for installing the CPU are too small for standard screwdrivers. I had to go digging through my toolbox to find a smaller gauge screwdriver to install. According to Sonnet, you must (though first hand experience says the contrary) also run a firmware patch to get the CPU to be recognized and take advantage of all the technology. This required multiple reboots and was more a nuisance than anything else. Reliability may be an issue. During the iTunes AIFF to AAC test the process actually failed the first time it ran. I immediately tried again and it finished without issue, but I was not expecting a failure on a simple task of processing music into a iTunes library. 
I was happy with the performance increase in Photoshop, especially with the re-drawing of previews while working with different filters. It did a good job of improving the time it took to do filters and was fast enough to not have to walk away whenever applying a filter. Still, it swasn't fast enough to keep up with me. This would be a great upgrade for a home user as it improves overall function of your Mac, but it is not up to par for heavy users who want to really bang on their beloved Power Mac G4s. FastMac 1.47 GHz First and foremost, FastMac's design receives high marks. The card had the smallest footprint and best use of space while still retaining a nice look. It should also be noted that I didn't have to apply a single patch to get the card up and running. Once you were at the latest firmware, you could plug and play your CPU upgrade. That was a big bonus in my book. 
The FastMac upgrade is only 170 MHz faster than the Sonnet and its price per Gig was very close to the Sonnet $177 per Gig. The improvements to the overall system were similar to the Sonnet card but there were times when the extra power was noticeable. Basic OS performance differences were equal to the Sonnet, but full screen movies ran even smoother, took less time to load. Also, file loading was dramatically faster than I was used to. While it was consistently just a step faster than the 1.33 GHz Sonnet card, the big plus came in stability. Everything seemed to be a little more stable on the FastMac card than on the Sonnet, and far faster than Apple's stock 733 MHz. 
The FastMac CPU upgrade started to show the limits of the rest of my system as it strived to keep up with the upgrade. The limits of the video card and the PCI bus started to take its toll on the performance. For example, the Photoshop test did not have as big of an improvement as I would have thought for the Actions test, but did a great job on the Radial blur test. To me this card is worth the extra twenty dollars to get the added performance reliability. 
Newer Technology 2.0 GHz Newer Technology's recent line of G4 upgrade cards really perform. Overall, this was the fastest of all the cards and pushed my Mac to the absolute limit. The performance increase throughout was amazing, especially in Photoshop. Every test, for the exception of the UI test, was handled with ease. Though, like the FastMac upgrade, my system probably slowed down the full capabilities of the upgrade. This card cost the most, priced just under $400, with a cost per Gig of $200--about $25 more per Gig than the competition. The installation of the required software upgrades was easier than the Sonnet card, as a bootable CD came with all the special upgrades on it. 
The design of the card came from a different direction than the other cards. Rather than have an open heatsink, the OWC card had a low profile and abundant fan power to push a large amount of air over an "enclosed" heatsink. I applaud Newer Technology for trying a different direction with their design. Almost every aspect of the card outperformed the other cards. 
Again, I stress this card clearly showed me the limits of my Mac. While almost everything ran faster on this card, the real world translations were not always in relation to the numbers. With a overall processor speed increase of 273% from the stock system, the average actual speed increase was around 210%. This difference is not related to the processor, but more related to the other components of the system. Most noticeable was the lack of increase to the frames per second on iTunes. Even though there was an extra 500 MHz of power, my little old Mac could only pump out an additional 1 frame per second using the 64MB NVIDIA GeForce MX card. The performance of this card would be better appreciated in a later model G4, as the improved system components would be able to take better advantage of the power. Now don't get me wrong, I loved this card. Everything went fast, I was literally in shock to see Photoshop and iTunes performing at such a high speed. This card is worth it if you want to keep you current box, but need a major power upgrade. 
The fans run off of a standard internal power cable, so you have to have a non-used power cable to run the card. If you are running multiple hard drives in your case, be weary of this aspect. While the extra fan noise did not bother me, there were more issues when dealing with this upgrade. While the software upgrades required to run the processor were simple to install, they were numerous and not permanent. You have to keep the CD around as some settings are stored in the NVRAM, and a zap to the PRAM would cause those settings to be erased, thus making the card inaccessible. The one benefit to the Newer Technology CD was how easy it was to use. Boot to the CD and follow a few terminal prompts and you can do all the upgrades and a single reboot. The oddest thing about this upgrade card was the performance in the UI bench mark. The XBench UI test showed a noticeable decrease technically, but could not be detected in real life performance. I'm not sure if one would notice the difference in the user interface refresh rate between Newer Tech's 180 FPS and the average 230 FPS rate in FastMac's and Sonnet's. However, I can say that the increase was far better than the stock 86 FPS. I did not let the UI test sway me against the card, as the actual performance did not follow the numbers this time. Real World Tests These tests were done using a stopwatch, my eyes and a few programs you can download on the 'net. Some of these are not "scientific" standard tests, but they are what I really saw while running these cards through the loops. Like most Mac modders out there, what is technically feasible is not important to me. With that said, here's what I found. Photoshop The first test I gave was a twenty-one step Photoshop action that employs rotations, blurs, unsharp masks, color conversions, filters and image reduction. This can be found on Other World Computing here. This results are given by displaying how many seconds it took to complete the steps. Interesting enough, OWC claims that my Mac should have finished 72 seconds earlier than it did running the stock 733 MHz processor. However, they also have their processor taking an additional 37 seconds over what my tests reported. These differences may be because of the version of Photohsop as I tested on CS 2 and they on CS. However, as you can see all processor upgrades gave a dramatic improvement in time. This just comes down to the extra processing power that these cards bring with them. 
The second Photoshop test was done to a 1500 x 2000 PSD document. I applied a radial blur to an edited, flattened file at the following settings : Amount - 100, Method - Spin, Quality - Best (see screen shot). This filter takes a good amount of time to complete and is focused to one aspect program. The results fell where one would expect them to. This was a good example of the difference in performance between the cards. As the chart shows, all times were decreased over the stock processor and, for the most part, the increased performance was in ratio to the processor speed. The question that is raised is how much time do you want to save? The OWC card clearly shows that highest time saving, but more interesting is the difference between the FastMac and the Sonnet offerings. The extra 11 seconds may not seem like much, but we all know stuff like this can add up. It is your own judgement as to how important that time difference is between these two cards. Shared Computing Rating The next test was using Boinc! for shared computer processing power. The processors are sent simple Integer and Floating point test and come back with how well they handled the test. All test came back in ratio of what their advertised processing power can handle. for example, the difference between FastMac and OWC is an approximate 136% increase, with the Boinc! ratio reading the same). The most interesting aspect of this test was the ratio between the stock CPU and the upgrades. The floating point test came back reading the correct ratio, but the integer ratio came back at 7% and 10 %better on all upgrade cards. The second chart shows the noted improvement in the upgrade cards over the stock card; each upgrade card has just slightly better ratios than the stock card. Logically, this and the xBench CPU test shoed me that the processors were performing correctly but my system was having a hard time keeping up. 
iTunes Next, I tested iTunes. This is a very common program and would really allow me to see what these processors could do in our daily activities, like importing and ripping AIFF downloaded CD to AAC format. The CD I chose was approximately 663 MB in AIFF format. I downloaded the music to my hard drive to bypass any CD reading issues. This test really showed the value in the improved processors. All upgrade cards decreased the time from "get up and get a cup of water and watch some TV" to "open a web page and read an article". The main issue that came out of this test was the first stability issue with Sonnet. It crashed on an import. This was the exact same imports that I had done on the other processors without any issues. The second test in iTunes was with the visualizer. I played "Linus & Lucy" as an AIFF, on medium size, using full screen : Radar Sweep, Dandilion Psyc, Hue Wheel settings. While this was running, I watched the frames per second and recorded what the frame rate was most stable at. The biggest eye opener for this test was the lack of increase between the upgrade cards. While there was a huge increase from stock performance, the upgarde cards stayed at 34 and 35 frames per second. This test showed that even with a faster processor card, it is not the "be all and end all" of upgrades. From what I can demise the video card could not handle any more operations. xBench The final say for me in numbers has always been XBench. Running XBench a few times and averaging the score usually shows the best way to quantify the quality of the individual pieces of hardware. However, this is not always the case. The two exceptions to the rule came up in my testing, the UI and Open GL. First, UI speed actually went down the faster the processor went. While it did not have a noticeable effect on running the OS or any other part of the software, I was surprised to see it. Again, cause of the slow-down may be associated with the rest of the hardware. No matter how fast of a CPU I put into my Mac, there are still components that will act as a bottleneck to the CPU's power. The second most noticeable difference in the xBench results was the Open GL test. I was expecting a much larger improvement in the test, but it came back showing only nominal improvements on the Newer Technology card (nominal compared to the difference in processor power). While technically the Newer Technology card was 36% and 50% faster than the Fast Mac and Sonnet offerings respectively, the improvements in the Open GL test was only 13% and 25%. Not the difference I was expecting. Though the Open GL and UI test showed some flaws in my system with these upgrade cards, I still considered it reliable due to the other test results. It clearly showed that the processors themselves can run at the advertised speeds advertised and that the improvements are measurable. Final Say In all honesty, it was harder to get the firmware updated than it was to put the new hardware in. All cards required a firmware update to 4.28 in order for them to run. Concordantly, all of three of these cards are great cards. The Sonnet card should attract any budget minded user with a lower/earlier G4 model. You will love the speed improvements and how it will allow you to keep your G4 viable a little longer. Though there were a few hiccups, the performance was still good and it was the most inexpensive of the models tested. If you have a little bit more money to throw at an upgrade, or if you are running a mid-model G4, then you should seriously consider FastMac. The extra performance is easily worth the $20. For people who don't want to go through the hassle of adding extra software components, FastMac is the way you should go. The Newer Technology/OWC upgrade should be for people with a higher end G4 that want to take a single processor G4 to the limit. The performance was awesome, but the rest of your Mac will slow you down if you don't upgrade your other components. Also, beware of the need to keep the CD around.
Write Comment |
||||||||||
| < Prev | Next > |
|---|
|
|