Faster Flash Memory Makes a Big Difference, More RAM or a Faster CPU?, OS X 10.4.10 Problems, and More

• Faster Compact Flash Makes a Big Difference
• More RAM or a Faster CPU?
• Level 2 vs. Level 3 Cache
• Problems after Upgrading to OS X 10.4.10
• Correction on 64-bitness of OS X
• Road Apple Nomination: 24" iMac
• LEM Unbalanced on Mac Gaming

Faster Compact Flash Makes a Big Difference

From Bob Casper following up on Problem with Addonics Adapter and CompactFlash:

Hi Dan,

Thank you for the information. That does make sense when the bus speed and flash speed are compared directly. The system RAM is maxed, at 576 MB to take advantage of the virtual memory. I returned my 16 GB 133x flash card and replaced it with a Transcend 8 GB 266x flash card (40 MBps).

The result is a very quick and responsive iBook for most tasks. There are some things I can't do on the Web, like play streaming videos. I have noticed my battery lasts longer on a full charge now, going from about 4 hours to well over 6 hours. And of course it's very quiet! I went from 16 GB of storage down to 8 GB, but at least it's more than my original hard drive, which only had 6 GB. I'm very happy with my upgrade.

Thanks again for the advice.

Bob

Bob,

Thanks for the report. I'm glad to hear the faster Compact Flash has improved things so much.

Dan

More RAM or a Faster CPU?

From Rhoda Stelchek:

Which is better/faster, a 2.0 GHz Mac Pro w/2 GB RAM or a 2.66 GHz Mac Pro w/1 GB RAM? Why? There will be no money in the foreseeable future to buy more RAM. This would make a good subject for an article as many others have to make a similar choice. Thank you for your time and consideration.

Rhoda Stelchek

Rhoda,

It depends. If you run a lot of applications, if you use virtualization, or if you use software that isn't universal binary, the additional memory will generally be a bigger benefit than the 30% difference in CPU speed. If you only run universal binary software and don't use virtualization, you'll benefit more from the additional processor speed.

Dan

Level 2 vs. Level 3 Cache

From Scott Cook:

Dan Palka said:

"On the Macs mentioned, the L2 cache is a full-speed 256 KB L2 cache, with a 2 MB half-or-less-speed L3 cache. If you replace it with an equal-speed processor that has a 1 MB L2 cache, but no L3 cache, it will be faster. The significant increase in L2 cache size will cause it."

Is this really true? I have always read that the presence of L3 is the most important factor, followed by the size of L3, followed by the speed of L3. I have often read that you can just add the L2 and L3 together when considering cache size. I believe there are 1.4 GHz G4 upgrade processors with and without L3 currently available. As I recall, the 7455 is faster because it has 2 MB of L3, even though L3 is slower than L2. It stands to reason that you can't fit as much in a 1 MB cache as you can in a 2 MB+ cache. Consider that the next step beyond cache is RAM, which isn't as fast. Anything that won't fit in the 1 MB L2 cache will be in RAM, which is slower than cache. Perhaps I have been reading misinformation? If so I apologize for spreading it around even more. I am neither a certified computer technician nor a zealot! (laugh)

I will keep reading your website because it makes me think, and it's very educational. Please keep up the good work... (smile)

Scott Cook  

Scott,

Everything is relative. Level 1 cache is part of the CPU itself, so it's accessed almost instantly. Level 2 cache isn't technically part of the CPU, and it can reside on the system board, between the CPU and the system board ("backside" cache), or as part of the same package as the CPU. In the last case, it's accessed almost as fast as L1 cache, as the CPU can access it as soon as it determined that the requested data isn't in the L1 cache.

If the L2 cache sits between the CPU and the motherboard, it will generally be accessed at 1/3 to 1/2 of CPU speed. Thus there's a greater delay in polling the backside cache and receiving data from it. And if the L2 cache is on the motherboard, it's constrained by the speed of the memory bus, which can be less than 1/10 of CPU speed in some cases.

There are several factors at work here: The larger the cache, the more likely it will hold the data you seek. The faster the cache, the more quickly the CPU can access it. It's not quite an even tradeoff: doubling cache size doubles the odds your data will be there. And a 256 KB onboard L2 cache will access data faster overall than a half-speed backside cache that is twice as likely to have the data you seek. There's just more latency the further the cache is from the CPU.

As you note, system RAM is the slowest RAM of all, so there is some benefit to having an L3 cache.

I'm sure there is better information on this subject on the Web, but the gist of it is that a twice as fast, half as large cache tends to outperform a slower, larger cache.

Dan

I just read this on Bare Feats:

'Another often asked question is, "Should I get CPU upgrade with the fastest clock speed or go for a slower clock speed version with L3 cache?" The short answer: "It depends."

It depends on what applications you are running. In past testing, certain applications (such as Photoshop) thrived on L3 cache. A slower CPU with L3 cache (7455 or 7457) would often beat a faster CPU with only L2 cache (7447A) when running our test action sets. But this discussion has recently become academic since the upgrades with L3 cache are being phased out. In a very short time, they will no longer be available unless leftover in some resellers inventory.

Also, in our most recent tests, the fastest 7455 (1.42 GHz with L3 cache) could not match the speed of the fastest 7447A (2.0 GHz with L2 cache). That and the fact that 7447As generate less heat and use less power, we have to recommend the 7447A based upgrades like the one we tested in this article.

Then again, if you can get a good price on a G4/1.4 GHz 7455 or 7457 based upgrade, it would give a G4/1.8 GHz 7447A a close race on all our test apps and beat it running Photoshop.'

Scott

Problems after Upgrading to OS X 10.4.10

From Phil:

Hello Dan,

Your site is great; I read it every day after I do that 'voodoo magic' with relational databases (as my coworkers call it) and then sit and wait for someone's computer to break. Yes, my company uses Windows, so I'm not idle too often.

I have an iMac, 1.83 GHz, 1.5 GB of RAM, and had been happily running OS X 10.4.9 for many months. About a week ago I decided I 'needed' an iPod touch. Needless to say, Capital One was more than happy to fund my whim. When I got home and ripped off the shrink wrap and plugged it in, I was told I had to upgrade to 10.4.10. I did so with trepidation, but no new updates have been issued, and .10 hasn't been recalled, how bad can it be?

Well, a few days after the upgrade and euphoria of transferring all my music, strange things started to happen. I don't remember the first few things that happened, but after a few days on 10.4.10, I woke up one morning and tried to check my email, no luck. Safari would open and do absolutely nothing. The wireless signal was fine, so I blamed it all on Comcast and unplugged the modem and router and went about my routine. I plugged the router and modem in a few minutes later, but the situation on the iMac was the same.

I went to work disgruntled; however, by the time I arrived at work, my Mac had pulled all my mail into the Apple Mail client and all the messages on webmail had been moved to read status. The computer is supposed to sleep after an hour; it stayed on all day. It has maintained this habit for the past five days. I was irritated but thought, hey, Leopard should be out soon, I'll just hold off on a reformat.

Fast forward to today. I can no longer open iChat, Mail, or iTunes. They bounce once in the dock and then stop: no load, no error message. I repaired permissions, performed the extended hardware test, pulled all the RAM and tried using each chip individually, created a new user account, ditched the Bluetooth keyboard and mouse, and restarted, all to no avail.

I'm a grad student with a procrastination habit, and I have a big paper due soon. I bit the bullet and reformatted (oh how I wish I'd been backing up so I wouldn't have had to throw everything I thought I needed on to an external hard drive). I can't blame all of these problems on 10.4.10, but I certainly didn't do anything else out of the ordinary. Obviously this email has no technical expertise, but I was just curious if anyone else had extraordinary problems after upgrading to 10.4.10.

Thanks,
Phil

Phil,

I can't say that I've been happy with 10.4.10 either. The first thing I noticed, several days after the upgrade, was that I could no longer print to my Brother HL-5250DL laser printer over the network. My Mac sees it as a Bonjour device, but that's as far as it goes. Fortunately it's close enough to my monitor (which has a built-in USB 2.0 hub) that I can use a 6' USB cable.

I've also been having completely weird, inexplicable icon problems. The longer my user session is open, the more apps I use, the longer I work, the more likely I am to see icons appear damaged. I've turned off several programs that load when I log on, but the problem persists. All I have to do to fix it is log out and back in again. Very weird.

I'm a firm believer in backups and testing upgrades, and some of the oddities I've seen in 10.4.10 didn't manifest right away. I should have stayed at 10.4.9, but field reports on the latest update were so positive I finally decided to do it. (I have since done a clean Tiger install and update, and the problem persists.)

I share your frustration. A patch update like this should make things better, not worse. Here's hoping for a 10.4.11 release that will fix the problems introduced with 10.4.10.

Dan

Correction on 64-bitness of OS X

From Yuhong Bao:

Dear Dan:

In OS X 10.5 and Mirror Drive Door Power Macs you wrote:

"There are costs to being an early adopter, and Intel's failure to make the Core Solo and Duo 64-bit processors was one of them. That said, the Mac OS has never been a 64-bit OS, although Tiger does have a little 64-bitness about it (it can access more than 2 GB of RAM, but no program can access more than 2 GB)."

should be: "There are costs to being an early adopter, and Intel's failure to make the Core Solo and Duo 64-bit processors was one of them. That said, the Mac OS has never been a fully 64-bit OS, although versions of Mac OS X since the G5 version of Mac OS X 10.2.7 does have a little 64-bitness about it (it can access more than 4 GB of RAM, but no individual program can address more than 4 GB) and Tiger supports 64-bit command line applications."

Yuhong Bao

Yuhong Bao,

Thanks for the correction. I've updated the article.

Dan

Road Apple Nomination: 24" iMac

From Benjamin Hays:

Hello Dan

I am a long time reader to your website. I read Anonymous' letter about Road Apples, and I'll like to nominate another one: The 24" iMac Core 2 Duo (Late 2006).

According to Engadget, this model features a "Mobile PCI Express Module" (MXM). However, the MXM Upgrade Store states: "The 24" iMac uses a card that is based on the MXM form factor but offers no compatibility. A reader of MXM Upgrade has tried several MXM cards in his Apple, including a reference 6200 card MXM Upgrade has send him, and none of those worked. If we are correctly informed, this comes from an incompatible vBios and this behavior was also seen in the past with incompatible AGP cards." (You'll need to scroll down to find it.)

Your definition of a "Road Apple" is a computer that is "deliberately compromised for no good reason (or no reason other than marketing)". I believe this iMac fits that definition of a "Road Apple".

What do you think?

Benjamin

Benjamin,

If Apple had promoted the 24" iMac (Late 2006) as having user upgradable graphics, this might have been worth noting on the computer's profile, but Apple has categorically stated that just because they use an MXM graphics processor doesn't mean that replacing it is supported. Quite the opposite, Apple says that it is not considered a user upgradable part.

Dan

LEM Unbalanced on Mac Gaming

From Mike:

Dan,

First of all, I have to clarify that I enjoy reading and exploring the Low End Mac site almost daily. Most of the time, I agree with what you and your columnists have to say. That said, I could not help but respond to some of the statements recently posted in under the heading of "Playing Games Not the Measure of Computer Power."

Most of the articles and responses I read on the site are well-balanced, but this thread seems to off-handedly disregard the gaming community. This is an industry that moves many millions of dollars a year. These people aren't the stereotypical losers living with their parents - they are those successful enough to afford the equipment you mention. Why is it such a bad idea to have Apple focus on bringing some of that cash into its collective pocket?

As you can guess, I'm a gamer. I long for the day that Apple will let me spend some extra cash to get a very high-end graphics card slapped into a Mac Pro to power both Aperture and the most advanced games the industry puts out. It is this antiquated view of Macs for one job and PCs for another that hurt Apple's potential growth.

Mike

Mike,

Thanks for writing. I'm not opposed to gaming on Macs, and I would love to see that area take off. Macs are the best all around computers on the planet, thanks to OS X, but there are areas where Macs aren't the best solution, whether that's due to hardware limitations (in the case of gaming, video cards), limited software selection, or cost.

I'm not a gamer, but even gamers agree that the Mac is not great for gaming. Inside Mac Games' Tuncer recently wrote:

Let's not mince facts here . . . the Mac is NOT a great gaming platform. There are a lot of drawbacks to it. It's a rather small market. Apple has one or two people at the company concentrating on Mac games (which is pathetic). The consoles are leaving computer gaming in the dust.

You may think it "antiquated" to view the Mac as a productivity platform and recognize that there are better platforms for playing, but most people see that as the reality. That said, the Mac is a decent gaming platform with a decent selection of games available.

For most of the past decade the Mac has been a minority platform with about 3% market share. Over that period we've seen a lot more change than the PC world: Macs moved from G3 (introduced just 10 years ago) to G4 to G5, then embraced x86 with Core Duo and now Core 2 Duo. And we went from Mac OS 7.6 through 8.x to 9.2.x and the huge transition to OS X starting in 2001. In other areas, we've gone from PCI graphics to AGP 2x, 4x, and 8x, and now on to PCI Express. And it was about 10 years ago that Apple finished the switch from SCSI hard drives in top-end Macs to EIDE drives.

The Mac has been a moving target to a much greater extent than Windows, where the x86 architecture has been used since 1981, PCI slots have been used since 1993, and AGP video since 1997. Windows itself has developed without the kind of discontinuity we've experienced with the launch of OS X and later with the switch to Intel x86 architecture.

Given money to spend on game development, are you going to write for 95% of the market or 5%? Are you going to target the stable platform or the one that keeps changing? Are you going to go where the gamers already are or are you going to either get Mac users to embrace gaming or give the Mac such a good game that Windows users will want to switch?

It's an uphill battle, and it's going to take a lot more than a super duper graphics card as a built-to-order option to make the Mac a real contender. It's going to take programmers developing games that work well on all levels of Macs, not just the high end. The higher the hardware requirements, the smaller the market, the less money to be made.

The odd thing is that some of the most fun computer games ever were developed for the old compact black & white Macs - Dark Castle, Shufflepuck Cafe, and Crystal Quest among them.

Maybe the solution isn't creating yet another 3D massively multiplayer role playing game, of which there seems to be no end, but dreaming up the kind of games that are fun to play but not nearly as demanding of system resources.

Think Different.

Dan