PowerCache Card in IIci -- Why So Slow?

retr01

Senior Tinkerer
Jun 6, 2022
2,473
1
793
113
Utah, USA
retr01.com
IIsi RAM-Muncher INIT

I would think it could work on a IIci but probably only if you have 1MB of RAM in Bank A.

So, if you have more than 1 MB in bank A, this INIT would not work? Hmmm. Isn't that something that can take advantage of to create a new INIT that can do the workarounds as discussed to get the IIci process better?
 

Fizzbinn

Tinkerer
Nov 29, 2021
189
183
43
Charlottesville, VA
So, if you have more than 1 MB in bank A, this INIT would not work? Hmmm. Isn't that something that can take advantage of to create a new INIT that can do the workarounds as discussed to get the IIci process better?

Well the IIsi which this Extension is specifically for has a fixed 1MB soldered on the Logic board for Bank A. I imagine the author would have just hard code it to eat up the rest of that 1MB not used by the RAM based built-in video. I suppose if there is a way to code to detect how much memory is actually in Bank A and then eat up that amount of RAM the Extension might be able to be dynamic in that way, usable in the IIci with different amounts of RAM in Bank A, but that’s above my knowledge (basically zero) of Mac programming to know if that could have been done back then for this Extension or if someone could develop a new one to do so. Maybe a Control Panel that lets you set an amount of RAM to "Munch" would be easier...

By the way there is better explanation of the CPU slow down issue with using RAM based built-in video (RBV) on page 46 of the IIci hardware Deve note:

The RBV and Bank A of DRAM share a separate RAM data bus, which can be connected to or disconnected from the CPU data bus by F245 bus buffers (see “Use of RAM by Video” in Chapter 3, “The RAM Interface”.) Data stored in Bank A of system DRAM is used by the RBV to feed a constant stream of video data to the display monitor during the live video portion of each horizontal screen line. The RBV asks the MDU for data as it is needed; the MDU responds by disconnecting the RAM data bus from the CPU data bus and performing an eight-long- word DMA burst read from RAM while clocking the read data into the RBV FIFO.​
If a video burst is in progress, a CPU access to RAM Bank A is delayed, effectively slowing down the CPU. This effect is more pronounced for the larger monitors and for more bits per pixel as you can see in Figure 5-1. Note that only accesses to RAM Bank A are affected by video. The optional Bank B of DRAM connects directly to the CPU data bus, and the CPU has full access to this bank at all times, as it does to ROM and the I/O devices.​
 
Last edited:

ScutBoy

Administrator
Staff member
Founder
Sep 2, 2021
337
316
63
Northfield, MN USA
So, if you have more than 1 MB in bank A, this INIT would not work? Hmmm. Isn't that something that can take advantage of to create a new INIT that can do the workarounds as discussed to get the IIci process better?

So - wrong name for the app and the wrong machine - about par for how my day has been going :)
 

joethezombie

Tinkerer
Oct 30, 2021
32
28
18
Idaho
I know we're discussing the PowerCache, but from the Turbo040 manual:

Some system extensions, such as A/ROSE, can disable the data cache and lower the performance of your system. If your Mac does not require the use of the A/ROSE software, we recommend that it be removed from your hard drive.

Again, not sure if that would apply to PowerCache, but certainly an avenue to explore.
 

Fizzbinn

Tinkerer
Nov 29, 2021
189
183
43
Charlottesville, VA
Awesome resource - thanks for posting! Are there more DevNotes for other models?

Oh yeah there are! I find them super interesting to read through. The "Architecture - System Block Diagrams" are particularly neat.

If you Google "<Mac_Model_Name> Apple Hardware Developer Note" it often results in a copy archived somewhere live today, alternatively you can get them off the Internet Achieve of Apple's Developer site, this one from 2003 lists developer notes for hardware back to the IIci:

 
Last edited:
  • Like
Reactions: retr01

Patrick

Tinkerer
Oct 26, 2021
434
1
224
43
speaking of super interesting things. in the classic devnotes

Do not attempt to connect 400 KB floppy disk drives to the external drive port; they are not supported by the Macintosh Classic computer and will not work. Earlier Macintosh computers provided a signal that controlled the disk speed of these drives; the Macintosh Classic, however, uses this signal in the power/sweep circuitry to control changes in screen brightness. ♦
 
  • Like
Reactions: RetroViator

phipli

Tinkerer
Sep 23, 2021
94
106
33
@This Does Not Compute

I just watched your video and it surprised me. I have the older vanilla PowerCache, and get the following performance in my IIci :

IIci PowerCache.jpg
 

phipli

Tinkerer
Sep 23, 2021
94
106
33
Further to the above, lowendmac has a speedometer 4 score for the IIfx CPU of 0.71. My IIci (using a Nubus video card and a 50MHz PowerCache) scores 0.865 for CPU, PR : 0.921.

I think some of the issue here is that the IIfx CPU scores are being assumed to be higher than they are. The PowerCache benchmarks higher than the IIfx for CPU in Speedometer and Norton System Info.

20230422_123411.jpg
 
Last edited:

phipli

Tinkerer
Sep 23, 2021
94
106
33
One more reply sorry about this, I've been playing with various benchmark software to see if the IIci scores differently on any of them out of curiosity - my copy of MacBench Speedometer 4 didn't have a score for a IIfx, so I ran a copy of 3.06. Here you can seem my 50MHz IIci vs the IIfx, and once again it shows the PowerCache winning vs the stock IIfx.

20230422_170740.jpg



Edit - whoops, that's Speedometer 3. I'll run an older version of MacBench shortly and add it here.

Edit 2 - MacBench 2 had a IIfx for comparison. It seems that the MacBench processor benchmarks score the IIfx proportionally higher than the IIci, vs the other benchmark software.

20230422_180511.jpg
 
Last edited:

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
I'm testing out a DayStar Universal PowerCache card in my IIci. It's the 50MHz 68030 version, so it *should* be wicked fast. And it's certainly faster than the IIci in stock form, but the benchmarks suggest it's not performing as it should.

The IIci has 20MB of RAM and is running System 7.0.1 with 32-bit addressing enabled. I have the PowerCache control panel installed, and both the cache and FPU are enabled, and it says the 50MHz 030 is active. The thing is, MacBench 2 says that the system (with card) only performs at 60% of what a IIfx does (that one has a 40MHz 030), though FPU scores are comparable. Speedometer rates the CPU at .74, while an entry for "50MHz SE/30" (there's no saved score for a IIfx) has a CPU rating of .87. (A stock IIci is rated at .40.) Some individual tests (whetstones, etc) are faster with the PowerCache, but not all.

For a card that was advertised as making a IIci faster than a IIfx, what gives? Is this a bus limitation in the IIci that the IIfx doesn't have, or does this card not benchmark well?

For reference, here are Speedometer 4.02 results from my SE/30. with a 40mhz PowerCache and a 50mhz PowerCache. The machine has 8mb RAM, I think these results were under Sys 7.5.5.

50mhz is on the left, 40mhz on the right. Both cards had the control panel installed, with both PowerCache and PowerMath turned on.

PXL_20230311_200821599 2.jpg
 

phipli

Tinkerer
Sep 23, 2021
94
106
33
For reference, here are Speedometer 4.02 results from my SE/30. with a 40mhz PowerCache and a 50mhz PowerCache. The machine has 8mb RAM, I think these results were under Sys 7.5.5.

50mhz is on the left, 40mhz on the right. Both cards had the control panel installed, with both PowerCache and PowerMath turned on.

View attachment 12064
Interesting, I figure, same as the IIfx vs the IIci, the benchmark must completely fit within the cache because the RAM and system bus is a different speed on all of these machines.
 

This Does Not Compute

Administrator
Staff member
Oct 27, 2021
283
429
63
www.youtube.com
  • Like
Reactions: pfuentes69

pfuentes69

Active Tinkerer
Oct 27, 2021
380
293
63
Switzerland
For reference, here are Speedometer 4.02 results from my SE/30. with a 40mhz PowerCache and a 50mhz PowerCache. The machine has 8mb RAM, I think these results were under Sys 7.5.5.

50mhz is on the left, 40mhz on the right. Both cards had the control panel installed, with both PowerCache and PowerMath turned on.

View attachment 12064
Hello. I'm curious about the usage in the SE/30

What was the performance increase compared on the base system?

Did you get the PowerCache with the original adapter for the SE/30 or did you manage to get one?
I'm thinking on installing this in the SE/30 instead of the IIci and I'm trying to figure out about the adapter...
 

phipli

Tinkerer
Sep 23, 2021
94
106
33

JeffC

Tinkerer
Sep 26, 2021
122
79
28
Seattle, WA
Hello. I'm curious about the usage in the SE/30

What was the performance increase compared on the base system?

Did you get the PowerCache with the original adapter for the SE/30 or did you manage to get one?
I'm thinking on installing this in the SE/30 instead of the IIci and I'm trying to figure out about the adapter...

Unfortunately I don't have a pic of the accelerator compared to stock, and somehow my Speedometer records file became corrupted, so I don't have easy access to that info without re-running the tests. The increase was significant though. The pic below is from an accelerator benchmark thread on another forum, it is not my machine or my photo. You can see the comparison of a Diimo 50mhz card with FPU and cache vs. a stock SE/30. If you compare my 50mhz numbers from my post above with these Diimo 50mhz numbers you can see they are very similar, except for the disk, which depends on what type of disk you are using. So this pic is probably a good comparison of my 50mhz PowerCache to a stock SE/30.

My 40mhz card and adapter both came in an SE/30 I purchased.

There is a recent thread on another forum addressing the slow speed using this card in an IIci. Someone did a bunch of testing with different variables, and finally found the correct combination that made the machine benchmark nearly as fast as an IIfx. I believe the main bottleneck was using the onboard video, since the video card and CPU share the same RAM. Using a video card or setting the disk cache so the system and video were not competing for the RAM caused a significant performance increase. Do a google search for "DayStar PowerCache 50MHz IIci Performance Results"

1691867876054.png
 

JDW

Administrator
Staff member
Founder
Sep 2, 2021
1,547
1,353
113
53
Japan
youtube.com
Colin, I am sorry I do not have a IIci to test, but perhaps this feedback will be semi-helpful...

My SE/30 benchmark results in Speedometer 4.01 using the 50MHz socketed Daystar PowerCache (with Power Central v2.2 loaded at startup) tests similarly to what @JeffC reported...

1691977797893.png

I generated those results with the stock SE/30 ROM, 128MB RAM, booted into System 7.1.

OS version is very important when benchmarking. I almost always get slower results when booted into S7.5.5 versus S7.1, for example. System 6.0.8 is also great for testing, but you can't run the newest versions of many benchmarking apps under System 6.

When benchmarking with Speedometer, it's important to always report the version number because benchmark results were adjusted across versions. Version 4 will report different numbers than version 3.x.

I don't seem to have a record for the IIfx in my copy of Speedometer 4.01. And sadly, I just downloaded v4.02 from the Garden and found it doesn't have the IIfx either! Can one of you who have version 4.01 or 4.02 of Speedometer AND who also have the IIfx record please Export that individual record and post it somewhere for download?