Accelerate Your Mac! - Review of the MetaBox JoeCard G4 450MHz 2MB cache CPU upgrade

A Click shows your site support to my Sponsors

The Source for Mac Performance News and Reviews

Review: MetaBox's JoeCard^TM Z G4/450 2MB Cache
First Shipping 2MB G4 CPU Upgrade
By Mike
Published: 9/15/2000
(Updated 9/16/2000 for Altivec Fractal peformance w/later appl. version)

MacBench 5.0 and Memory Benchmark Test Results

Intro | Benchmarks | Appl. Tests | Software Controls | Installation | Specs/Design | Summary

Benchmark Tests

Benchmark tests were run with MacBench 5.0 as it is the accepted Mac standard. I've also included results with CineBench 2000 v1.0, an Altivec Fractal Generator, G4Timedemo and memory bandwidth test results.

Remember that Macbench does not take advantage of Altivec instructions (aka the "Velocity Engine" as Apple calls it). Also note that Graphics scores will vary depending on resolution, color depth and the installed video card. This is not a reflection on the CPU card under review. Ditto for disk scores, as the hard drive type, speed and free space/fragmentation all affect scores. The baseline 1000 score is based on the results with a Beige G3/300/150/1MB (with extensions on of course).

MacBench 5.0 Tests: The graph below compares the following configurations and systems

A B&W G3/400 baseline system.

A Blue and White G3 with the following CPU upgrades installed:

A G3/450 1MB cache (Formac ProG3/450)
The MetaBox G4/450 2MB cache upgrade (with OS 9 & Altivec extensions)
Both these upgrades were tested with the original ATI Rage128 graphics card and a Voodoo5 (noted as ATI and V5 in the graph).

An Apple G4/500 AGP (single CPU, rev 2.8)

The B&W G3 were identical except for the MetaBox enabler and OS 9 Altivec extensions and any Video card drivers required for ATI or Voodoo5 card. OS version, hard disk, installed RAM, etc. were exactly the same for the B&W tests. The G4/500 AGP system ran the same OS version and as similar an extension set as possible. The bottom of the intro page lists each system's details.

Notes: Some explanation of the MacBench graph and test components.

Disk Scores: The B&W G3 tests used the onboard ATA/33 controller with a Maxtor 27GB ATA/66 drive (7200 RPM/2MB cache). The G4/AGP system used the original 20GB WD Expert ATA/66 drive (7200 RPM/2MB Cache) with the onboard ATA/66 controller. All disks used Apple's OS 9.04 drivers and were approximately 40% full and not optimized.

Graphics Scores: All tests were run at 1024x768, thousands colors and used a 66MHz PCI (B&W) or AGP slot (G4/500 AGP). The B&W G3 was tested with an OEM rev 2 ATI Rage128 using the latest drivers and with a 3dfx Voodoo5. The Voodoo5 used the current 1.01 drivers with 1.31 ROM. The G4/500 AGP system used a Rage128 Pro AGP card with the latest drivers. See my video cards page for reviews of alternative graphics cards for these systems.

Fractal FPU Benchmark:
(Note: A reader replied that there is a newer version than I used for the original results posted here. The version I used originally was last modified 5/16/2000 and reported appx. 990MFlops, the current download has a last mod date of 6/26/2000 and reports nearly 50% higher rates with a G4. I've retested with the later version and updated the graph below to reflect the results.)
I used a freeware Altivec Fractal demo program (available here) to compare FPU performance of the B&W system with a G3/450 CPU and with the MetaBox G4/450 upgrade. (Note: MegaFlops/sec = Millions of Floating Point Operations per Second.)

Fractal Demo Results
Support of Altivec instructions dramatically increases performance with the G4 upgrade versus a G3 in this test.

CineBench 2000 v1.0 Results:
I originally tested with the 1998 Cinebench and reported the results to Maxon, who said they were working on an updated version. Within days they released an updated Cinebench 2000 which I used for all the tests below. (For more info see www.maxon.net/pages/download/benchmarks.html.) For a full explaination of the benchmark and how to read the results, see this page.
All tests were run at 1024x768, millions color mode. Longer bars are faster.

CineBench 2000 Results
Note the disappointing results with the Voodoo5 as far as OpenGL mode performance. It was slower than Cinema's software mode. Software mode reported very slightly higher results with the Voodoo5 installed than with the Rage128. I ran CineBench 2000 on a friend's HP 1G Pentium III 1GHz with 64MB GeForce2 GTS; results that were 8.73 (Software), 14.21 (OpenGL) and 11.19 (Raytracing). I assume from these results CineBench does not take advantage of Altivec, but the faster PIII CPU and 4x AGP GeForce2 card also was a factor I'm sure. When my dual G4/500 with ATI Radeon 32MB DDR arrives, I'll be running CineBench tests on it to see how it compares.
I've added CineBench 2000 as one of the benchmark options on my Mac Game/Video card performance database. If you download CineBench 2000, please enter your results.

G4Timedemo Benchmark:

The graph shows G4Timedemo (v032) performance with B&W G3 using the MetaBox G4/450 (2MB cache), an XLR8 G4/400 and G4/450 (1MB cache), and a Formac ProG3 G3/450MHz (1MB cache) upgrade. My rev 1 G4/450 AGP system is also shown for comparison. Note the improvement a G4 makes over a G3 in this benchmark which takes advantage of Altivec/Velocity Engine instructions.

G4 timedemo Results

G4Timedemo is also one of the Benchmarks included in the Mac Game/Video card performance database. As shown on the entry page for the database, you can download G4timedemo at www.altorsys.com/HTMLAltor/ProjectB.html

Memory Bandwidth Tests:

Although 100MHz bus speed Macs have generally 100MB/sec or higher memory performance, note that as shown on the Applications Tests page, the huge advantage in memory bandwidth is often not a factor in many applications tests. The efficiency of the L1/L2 caches isolate the system bus speed in most cases to where bus speed is often not a factor in real world application performance.
Several readers have asked what effect a G4 CPU upgrade would have on memory bandwidth compared to a G3 CPU. The results below show the MetaBox G4/450 (2MB cache) vs. a G3/450 (1MB cache) using the same base system. Tests with the G4 CPU had the 4 Altivec OS 9.04 extensions active. I've heard that OS 9.04 has some Altivec support in the memory manager and the G4 did deliver higher memory bandwidth performance, even for file sizes larger than the cache (testing main memory performance).

Newer Tech's Gauge Pro 1.01 Results:
Newer Tech's Gauge Pro utility also includes a memory bandwidth test feature. For this review I used GaugePro v1.1b2. Some previous G4 reviews used an earlier version.

GaugePro Result w/MetaBox G4/450 2MB

The G4 upgrade GaugePro memory bandwidth scores were about 50% faster than the same system with a G3/450 upgrade installed, which reported 105.3MB/sec rates.

Stream Results:
The image below shows the results from the Stream benchmark with the G3 vs G4 CPUs:

Note the Stream results also show a benefit from the G4 CPU upgrade also. For comparison the image below shows Stream results with my G4/450 AGP (Sawtooth Maxbus motherboard). These results were with OS 9.0, and my G4/450 CPU was a rev 2.6 with errata.

Stream showed the Sawtooth having a significant advantage only in the Assignment test. The 2.6 rev G4 CPU (with disabled speculative processing)

Now the results of tests with another memory bandwidth benchmark that shows the effect of the 2MB cache and compares a G3 vs G4 CPU in the same system. Note how much faster the G4's L1 and L2 cache rates are compared to the G3 and even main memory bus rates got a boost from the G4 CPU. (Rates below the red line indicate data sizes larger than the backside cache, which show main memory bus rates.)

G4 2MB vs G3 1MB Results
Note the read rates for data sizes larger than the backside cache are about twice as high with the G4 CPU according to this benchmark, yet the write rates are about identical, with the G4 very slightly lower, by an insignificant amount (perhaps within the run/run variation).

Benchmark performance is interesting for technical reasons, but real world application performance is what really counts. Application and 3D Game tests are covered in the next page of this review. Or you may use the links below to jump to a specific page.

Index of MetaBox's JoeCard ^TM Z G4/450 2MB Cache Review
Intro | Benchmarks | Appl. Tests | Software Controls | Installation | Specs/Design | Summary

- or -
Back to WWW.XLR8YOURMAC.COM

Copyright © Mike, 2000.
All Rights Reserved.
All brand or product names mentioned here are properties of their respective companies.

Users of the web site must read and are bound by the terms and conditions of use.

The Source for Mac Performance News and Reviews
Review: MetaBox's JoeCard^TM Z G4/450 2MB Cache First Shipping 2MB G4 CPU Upgrade By Mike Published: 9/15/2000 (Updated 9/16/2000 for Altivec Fractal peformance w/later appl. version)
MacBench 5.0 and Memory Benchmark Test Results
Intro \| Benchmarks \| Appl. Tests \| Software Controls \| Installation \| Specs/Design \| Summary
Benchmark Tests
Benchmark tests were run with MacBench 5.0 as it is the accepted Mac standard. I've also included results with CineBench 2000 v1.0, an Altivec Fractal Generator, G4Timedemo and memory bandwidth test results. Remember that Macbench does not take advantage of Altivec instructions (aka the "Velocity Engine" as Apple calls it). Also note that Graphics scores will vary depending on resolution, color depth and the installed video card. This is not a reflection on the CPU card under review. Ditto for disk scores, as the hard drive type, speed and free space/fragmentation all affect scores. The baseline 1000 score is based on the results with a Beige G3/300/150/1MB (with extensions on of course). MacBench 5.0 Tests: The graph below compares the following configurations and systems A B&W G3/400 baseline system. A Blue and White G3 with the following CPU upgrades installed: A G3/450 1MB cache (Formac ProG3/450) The MetaBox G4/450 2MB cache upgrade (with OS 9 & Altivec extensions) Both these upgrades were tested with the original ATI Rage128 graphics card and a Voodoo5 (noted as ATI and V5 in the graph). An Apple G4/500 AGP (single CPU, rev 2.8) The B&W G3 were identical except for the MetaBox enabler and OS 9 Altivec extensions and any Video card drivers required for ATI or Voodoo5 card. OS version, hard disk, installed RAM, etc. were exactly the same for the B&W tests. The G4/500 AGP system ran the same OS version and as similar an extension set as possible. The bottom of the intro page lists each system's details. Notes: Some explanation of the MacBench graph and test components. Disk Scores: The B&W G3 tests used the onboard ATA/33 controller with a Maxtor 27GB ATA/66 drive (7200 RPM/2MB cache). The G4/AGP system used the original 20GB WD Expert ATA/66 drive (7200 RPM/2MB Cache) with the onboard ATA/66 controller. All disks used Apple's OS 9.04 drivers and were approximately 40% full and not optimized. Graphics Scores: All tests were run at 1024x768, thousands colors and used a 66MHz PCI (B&W) or AGP slot (G4/500 AGP). The B&W G3 was tested with an OEM rev 2 ATI Rage128 using the latest drivers and with a 3dfx Voodoo5. The Voodoo5 used the current 1.01 drivers with 1.31 ROM. The G4/500 AGP system used a Rage128 Pro AGP card with the latest drivers. See my video cards page for reviews of alternative graphics cards for these systems. Fractal FPU Benchmark: (Note: A reader replied that there is a newer version than I used for the original results posted here. The version I used originally was last modified 5/16/2000 and reported appx. 990MFlops, the current download has a last mod date of 6/26/2000 and reports nearly 50% higher rates with a G4. I've retested with the later version and updated the graph below to reflect the results.) I used a freeware Altivec Fractal demo program (available here) to compare FPU performance of the B&W system with a G3/450 CPU and with the MetaBox G4/450 upgrade. (Note: MegaFlops/sec = Millions of Floating Point Operations per Second.) Support of Altivec instructions dramatically increases performance with the G4 upgrade versus a G3 in this test. CineBench 2000 v1.0 Results: I originally tested with the 1998 Cinebench and reported the results to Maxon, who said they were working on an updated version. Within days they released an updated Cinebench 2000 which I used for all the tests below. (For more info see www.maxon.net/pages/download/benchmarks.html.) For a full explaination of the benchmark and how to read the results, see this page. All tests were run at 1024x768, millions color mode. Longer bars are faster. Note the disappointing results with the Voodoo5 as far as OpenGL mode performance. It was slower than Cinema's software mode. Software mode reported very slightly higher results with the Voodoo5 installed than with the Rage128. I ran CineBench 2000 on a friend's HP 1G Pentium III 1GHz with 64MB GeForce2 GTS; results that were 8.73 (Software), 14.21 (OpenGL) and 11.19 (Raytracing). I assume from these results CineBench does not take advantage of Altivec, but the faster PIII CPU and 4x AGP GeForce2 card also was a factor I'm sure. When my dual G4/500 with ATI Radeon 32MB DDR arrives, I'll be running CineBench tests on it to see how it compares. I've added CineBench 2000 as one of the benchmark options on my Mac Game/Video card performance database. If you download CineBench 2000, please enter your results. G4Timedemo Benchmark: The graph shows G4Timedemo (v032) performance with B&W G3 using the MetaBox G4/450 (2MB cache), an XLR8 G4/400 and G4/450 (1MB cache), and a Formac ProG3 G3/450MHz (1MB cache) upgrade. My rev 1 G4/450 AGP system is also shown for comparison. Note the improvement a G4 makes over a G3 in this benchmark which takes advantage of Altivec/Velocity Engine instructions. G4Timedemo is also one of the Benchmarks included in the Mac Game/Video card performance database. As shown on the entry page for the database, you can download G4timedemo at www.altorsys.com/HTMLAltor/ProjectB.html Memory Bandwidth Tests: Although 100MHz bus speed Macs have generally 100MB/sec or higher memory performance, note that as shown on the Applications Tests page, the huge advantage in memory bandwidth is often not a factor in many applications tests. The efficiency of the L1/L2 caches isolate the system bus speed in most cases to where bus speed is often not a factor in real world application performance. Several readers have asked what effect a G4 CPU upgrade would have on memory bandwidth compared to a G3 CPU. The results below show the MetaBox G4/450 (2MB cache) vs. a G3/450 (1MB cache) using the same base system. Tests with the G4 CPU had the 4 Altivec OS 9.04 extensions active. I've heard that OS 9.04 has some Altivec support in the memory manager and the G4 did deliver higher memory bandwidth performance, even for file sizes larger than the cache (testing main memory performance). Newer Tech's Gauge Pro 1.01 Results: Newer Tech's Gauge Pro utility also includes a memory bandwidth test feature. For this review I used GaugePro v1.1b2. Some previous G4 reviews used an earlier version. The G4 upgrade GaugePro memory bandwidth scores were about 50% faster than the same system with a G3/450 upgrade installed, which reported 105.3MB/sec rates. Stream Results: The image below shows the results from the Stream benchmark with the G3 vs G4 CPUs: Note the Stream results also show a benefit from the G4 CPU upgrade also. For comparison the image below shows Stream results with my G4/450 AGP (Sawtooth Maxbus motherboard). These results were with OS 9.0, and my G4/450 CPU was a rev 2.6 with errata. Stream showed the Sawtooth having a significant advantage only in the Assignment test. The 2.6 rev G4 CPU (with disabled speculative processing) Now the results of tests with another memory bandwidth benchmark that shows the effect of the 2MB cache and compares a G3 vs G4 CPU in the same system. Note how much faster the G4's L1 and L2 cache rates are compared to the G3 and even main memory bus rates got a boost from the G4 CPU. (Rates below the red line indicate data sizes larger than the backside cache, which show main memory bus rates.) Note the read rates for data sizes larger than the backside cache are about twice as high with the G4 CPU according to this benchmark, yet the write rates are about identical, with the G4 very slightly lower, by an insignificant amount (perhaps within the run/run variation).
Benchmark performance is interesting for technical reasons, but real world application performance is what really counts. Application and 3D Game tests are covered in the next page of this review. Or you may use the links below to jump to a specific page.

Index of MetaBox's JoeCard ^TM Z G4/450 2MB Cache Review Intro \| Benchmarks \| Appl. Tests \| Software Controls \| Installation \| Specs/Design \| Summary - or - Back to WWW.XLR8YOURMAC.COM
Copyright © Mike, 2000. All Rights Reserved. All brand or product names mentioned here are properties of their respective companies. Users of the web site must read and are bound by the terms and conditions of use.