Hardware

Name-brand memory vs. generic

Dave Farquhar Hardware , , , , , , , , ,

The difference between brand-name RAM and commodity RAM. I’ve been seeing a lot of questions along the lines of, “Do I really have to buy name-brand memory when I can find memory for half the price on PriceWatch?” on message boards lately. I talked about memory some in Memory-buying secrets, but I didn’t really go into the difference between generic/commodity/broker memory and the expensive stuff.

There are three factors that go into the quality of a memory module: the quality of the chips, the quality of the printed circuit board (PCB), and manufacturing.

When memory chips are made, they are tested. A memory chip that runs at or below spec gets classified as an A-grade chip. Chips with minor defects are classified as C-grade chips and shouldn’t be used in PCs.

Memory manufacturers will also charge varying amounts based on how much testing they do for the chips. Top-tier chips are guaranteed to have a failure rate of .1% or less–we’re talking one in a thousand chips failing here.

The least expensive chips aren’t tested at all.

The only way to ensure you’re getting these best-of-the-best chips is to buy name-brand memory. The best way to ensure you’re getting C-grade or untested chips is to buy the cheapest module you can find.

Inexpensive PCBs use a four-layer design, with signal layers on the outside, and power and ground sandwiched inside. This is inexpensive to produce and easy to repair. Unfortunately, this design leaves you open to signal noise, which can corrupt the data stored in the chip, and lead to unpleasant things like BSODs.

A better approach is to put the signal layers inside, and put power and ground on the outside, protecting them. Better still is a six-layer design, which adds two more ground layers for even better isolation. The higher the memory speed, the more important this extra isolation becomes.

You can sometimes tell the difference between a 4-layer and a 6-layer board by looking at it under a strong light. By turning it slowly, you can isolate discrete layers with the naked eye. However, it’s impossible to tell the difference between power, ground, and signal layers with the naked eye.

Name-brand vendors use high-quality PCBs. Some even proudly proclaim it when they use 6-layer boards. Again, the best way to get a poorly designed 4-layer board is to buy the cheapest memory you can find.

The third factor is assembly. When soldering chips to PCBs, things can go wrong. Trust me. I’m very good at demonstrating. While no one puts together memory modules by hand (I hope), my plumber-like soldering skills make me appreciate good equipment. Quality solder joints are bright and shiny, and they’re applied very quickly. Intermittent solder joints cause problems, and they’re maddening.

Kingston puts every module through rigorous testing. Other name-brand manufacturers test as well. When you don’t even know who made your module, it’s impossible to know whether it’s had the proper testing. Putting it in a PC and watching the BIOS check is not proper testing. Memory has to be very far gone to fail that test.

Miscellaneous computer memory.
The module on the right has Samsung chips on it, but that doesn’t necessarily mean it was made by Samsung. I’d be fine with putting the HP module directly below it in a PC, though. The same goes for the Kingston module at the top of the image.

Frequently vendors will advertise Micron memory or NEC memory. A Micron or NEC stamp on the chips doesn’t necessarily mean they manufactured the module! Be sure to find out who assembled the module–they usually stamp the PCB, or they put a sticker on the module itself. If you find a so-called Micron module on Pricewatch for half the price that Crucial is asking (Crucial is a division of Micron), chances are it’s a no-name module that just happens to have Micron chips on it.

So, does it really make a difference?

As an IT professional by day, I work on a large network–roughly 700 PCs. I’ll conservatively estimate that farm of PCs has 1,000 memory modules collectively. We buy name-brand memory (Crucial, Kingston, Viking or Simple) exclusively. We buy PCs from Micron, so they have Crucial modules in them. Macintoshes generally come with Crucial or IBM modules.

A typical memory module has 8 chips on it, and it only takes one bad chip to make the whole module bad. I’ve seen two bad memory modules cross my desk–so we’re talking two bad chips out of a batch of 8,000. So if you’re buying a single module for a home PC, your chances of getting a lemon are pretty slim if you get a good name brand.

For my own use, I buy name-brand memory modules. Usually this means Crucial; I got a great deal over the summer on some PNY memory so I bought a couple of sticks. I use high-quality memory, I don’t overclock, and I generally don’t run experimental software. I almost never get a BSOD or an illegal operation error, even when running Windows 95 or 98. I reboot my Windows 9x boxes about once a month on average, and half the time that’s because I installed or uninstalled something and it requires me to reboot. To give you an idea how I use PCs, at the moment I have seven applications running, with 11 windows open between them, and two TSRs running. That’s my idea of moderate use.

I’d rather have 64 megs of Crucial memory than a gig of the cheap stuff. Hopefully now you see why.

Upgrading by downgrading

Dave Farquhar Hardware

This is really sad. I don’t like complaining about supercheap video cards any more than you like reading about them, so yesterday I did something about it. I caught a glimpse of my components graveyard under my desk, and among the rubble I spied a Diamond Stealth 3D 2000 card. This was a mid-range card I purchased for $100 or so in mid-1997 to go with an Abit IT5H motherboard. For its day, it provided high color displays, high resolutions, GUI acceleration, and slight acceleration of some 3D-related functions.

We still have some of these in service at work and they’re adequate, so I figured it would be worth a try to swap it in for my AGP Cirrus Logic-based card. It only has 2 MB RAM on it, as opposed to the Cirrus’ 4 MB, but this system’s connected to a 15-inch monitor so that’s not an issue. I popped the hood, pulled the Cirrus, dropped the Stealth into an available PCI slot, powered on, and Win98 automatically installed S3 ViRGE 325 drivers. I’d have preferred to feed it the Diamond drivers I downloaded, but there was no harm in giving the MS-supplied drivers a shot. I let it reboot, then attempted the acid test. I loaded up Winamp, then IE. Once Winamp was running and playing, I went to The Register. Normally, when I scrolled in the browser, Winamp would screech. Not this time. Sweet silence, other than the music playing. Occasionally when viewing a particularly obnoxious site, I could coax a screech, but not consistently, and certainly not enough to be worth worrying about.

So, am I saying the mid-range chipsets of 1997 are better than the low-end chipsets of today from Cirrus and Trident, AGP notwithstanding? Yes I am, as much as it pains me. That’s not usually how technology works; for example, the low-end IDE hard disks of today kill even the high-end IDE disks of 1997. But I’m happy. That’s $75 I don’t have to spend on another TNT2 card.

Computer buying advice

Dave Farquhar Hardware , , , ,

Some sound computer buying advice. Here’s a Washington Post article on buying new PCs. Easy to understand in layman’s terms. And the advice is for the most part sound too, though I recommend always buying a good video card–a TNT2 will just add $60 or so to the cost of a low-end box and everything will run more nicely. The box I’m typing on right now has a cheap Cirrus Logic-based card in it, and the high CPU usage of its drivers hurts multitasking noticeably, even if I’m just browsing the Web while listening to music.

In a year this’ll be a moot point, as all chipsets will have serviceable embedded video. Even the enraging Intel i740, though not good for games, was great for productivity use and much better all around than this Cirrus and Trident garbage, and Intel’s newest chipsets have i740 derivatives in them. Future VIA chipsets will have S3 video in them. Same story.

I buy crap so you don’t have to–but don’t get me wrong. I buy the good stuff too. That way I’ll know the difference.

No more wimpy PC sound for me. I just connected an ancient but still awesome Harmon/Kardon 330A receiver (built in the late 1960s, I’m guessing — it once belonged to my dad) to my computer along with a pair of KLH 970A speakers I picked up for 30 bucks at Best Bait-n-Switch (unfortunately, the only nearby place that sells KLH speakers). These things are scarcely bigger than the cheap desktop speakers that came with the last PC I bought — 7 3/8″ high x 4 5/8″ wide by 4 3/8″ deep — but with the volume cranked to about 1/3 I can hear it throughout my apartment. I imagine at 2/3 I’d meet my neighbors. I won’t try that — I’m not interested in sharing my great tunage.

I can’t believe neither my mom nor my sister wanted this receiver — honestly, every time I’ve mentioned this thing at an audio place the salesperson has asked if I was interested in selling it — but hey, my dad would have wanted me to have a kickin’ audio setup for my PCs, right? This’ll work great for Royals broadcasts over the ‘Net once baseball season starts again, but not only that, this combination kicks out the jams almost as hard as punk legends The MC5, so I’m not complaining.

I’m happy enough with the results that I think rather than replacing my dying CD changer, once my Windows Me experiments are over I’ll mount my extra 15-gig drive somewhere on my LAN and put my Plextor Ultraplex CD-ROM drive to work ripping my entire CD collection, which I’ll then encode at 320 kbps. I doubt I’ll notice much difference.

If you’re like me and live with several PCs in close proximity to one another, rather than plugging an endless number of cheap desktop speakers into them, pick up an inexpensive receiver or use a castaway. You can plug a PC into any stereo input except phono, so most modern receivers should accomodate at least three PCs, and the speaker options are limited only by the receiver’s capabilities and available space. You’re likely to be much happier with such a setup than with any desktop speakers you’ll find, and a receiver plus speakers will usually cost much less than multiple pairs of any set of desktop speakers worth having would. Just be very careful to isolate your speakers away from any floppies and Zips and other magnetic media you might have. Some bookshelf speakers may be magnetically shielded, but don’t count on it.

Optimizing a K6-2

Dave Farquhar Hardware , , , , ,

Now I remember why I’m here. I was at my mom’s and stepdad’s, and of course the inevitable topic came up.
“I’ve got Kristin going through your book and implementing some of it,” John says. “Our computer is SLOW.” So you know what I did… I fired it up and gave it a look myself. (Besides, they couldn’t get Duke Nuke ‘Em to run and I’ve been working on an article on getting DOS games running inside Windows so I welcomed the chance to practice.)

The story: It’s an 18-month-old whitebox clone. AMD K6-2/350, 64 megs RAM. No idea what motherboard. Old non-UDMA 4.3-gig hard drive. Trident 9750-based video, and a C-Media ISA sound card. Oh yes, and of course, a Rockwell-based PCI softmodem.

A computer is a team, and what we’ve got here isn’t exactly the computing equivalent of the 1962 Mets, but I’d rate it about as highly as the 1996 Kansas City Royals. Making matters worse, it was loaded down with way too many gee-whiz features. I slimmed it down to what’s necessary, got its boot time down to well under a minute, and got the software running decently.
I’m guessing there are millions of PCs like this one out there. Buying a new one isn’t an answer, because if you buy one PC like this, chances are the next one you buy will have the same problems. You’ll probably get another softmodem. You’ll probably get another cheapo sound and video card. You’ll get a superfast CPU. But the problem is, that’s like adding the 1999 or 2000 edition of Mike Sweeney or Jermaine Dye to the 1996 Royals. You’ve still got a last-place team. The difference is, now you’ve got a last-place team with a .300 hitter or two in the heart of the lineup. Makes the team more fun to watch, and you’ll win a few more games, but the fundamental problem that got you into last place is still there.

Sure, I’d recommend some hardware upgrades for this beast. A SoundBlaster Live! sound card, for sure. An ISA sound card can eat up 35% of available CPU time. The SB Live! uses 1-2 percent. An inexpensive TNT2-based video card, certainly. At about $70, it’ll free up a lot of CPU time and speed the video along. I’d give serious thought to a newer, faster hard drive as well. A Maxtor 30-gigger (5400 rpm but still a massive improvement over this drive) runs $99 at CompUSA this week. Deals like that are common. And it wouldn’t hurt to get a controller-based modem as well. Zoom has such a beast available that’s getting good marks (I need to find the model number) for under $100.

Do all that, and you’re looking at about $350-$400 worth of upgrades. That’s a substantial chunk of a new computer, yes. But it re-uses the stuff that’s still serviceable under the hood. And these peripherals are all good enough to move on if you decide to replace what’s left (essentially, the motherboard).

So there’s a huge market for Optimizing Windows. People need this kind of information. Now, how to get it to them…? O’Reilly hasn’t figured out how to do it. It’s up to me.

Good thing there’s a marketing genius at work I can talk to. I’ll be bouncing this off him this week to see what he thinks. Then I get the first full week of December off. Time for a marketing blitz.

I’m still on the road. So if you’ve e-mailed me, it’ll be a day or two before I can get back to you. I’ll field all that when I get back.

Pentium 4 performance is precedented

Dave Farquhar Hardware , , , , ,

Thoughts on the Pentium 4 launch. No big surprises: a massively complex new processor design, limited availability, and systems from all the usual suspects, at high prices of course. And, as widely reported previously, disappointing performance.
This isn’t the first time this has happened. The Pentium Pro was a pretty lackluster performer too–it ran 32-bit software great, but Win9x was still the dominant OS at the time and it still has a lot of 16-bit code in it. So a 200 MHz Pentium Pro cost considerably more than a 200 MHz Pentium and for most of the people buying it, was significantly slower. History repeats itself…

Intel revised the Pentium Pro to create the Pentium II, with tweaks to improve 16-bit performance, but of course massive clock speed ramps made that largely irrelevant. Goose the architecture to 600 MHz and you’re going to blow away a 200 MHz previous-generation chip.

That’s what you’re going to see here. Intel fully intends to scale this chip beyond 2 GHz next year, and that’s when you’ll see this chip come into its own. Not before. And by then Intel will probably have changed their socket, (they intend to change it sometime next year) so buying a P4 today gives you no future-proofing anyway.

It never makes sense to be the first on the block with Intel’s newest chip. Never. Ever. Well, if you’re the only one on the block with a computer, then it’s OK. The P4 has issues. The P3 had issues (remember the serial number?) and was really just a warmed-over P2 anyway. The P2 was a warmed-over Pentium Pro. The Pentium Pro had serious performance issues. The Pentium had serious heat problems and it couldn’t do simple arithmetic (“Don’t divide, Intel inside!”). The last new Intel CPU whose only issue was high price was the 486, and that was in April 1989.

Unless you’re doing one of the few things the P4 really excels at (like encoding MP4 movies or high-end CAD), you’re much better off sticking with a P3 or an Athlon and sinking the extra money into more RAM or a faster hard drive. But chances are you already knew that.

Time to let the cat out of the bag. The top-secret project was to try to dual-boot WinME and Win98 (or some other earlier version) without special tools. But Win98’s DOS won’t run WinME, and WinME’s DOS seems to break Win98 (it loads, but Explorer GPFs on boot).

The best method I can come up with is to use the GPL boot manager XOSL. It just seems like more of an achievement to do it without third-party tools, but at least it’s a free third-party tool. You could also do it with LILO or with OS/2’s Boot Manager, but few people will have Boot Manager and LILO will require some serious hocus-pocus. Plus I imagine a lot of people will like XOSL’s eye candy and other gee-whiz features, though I really couldn’t care less, seeing as it’s a screen you look at for only a few seconds at boot time.

Fun with electricity

Dave Farquhar Hardware , , ,

Fun with electricity. I’m trying to figure out if I’m overreacting or not. What really scares me is that this journalist seems to know a whole lot more about electrical safety than some other people working in an IS/IT department.
The scenario: I had a PC that wouldn’t boot or power off. It sat there in a catatonic state, HD LED solid, power LED solid, fans running, but no other signs of life. The only way to power it off was to pull the plug. Plug it back in, and it reverted instantly to the catatonic state. I popped the hood and didn’t see anything obvious. I did notice a weird smell, which isn’t unusual for an electrical problem, but it was somehow different. Organic… I unplugged the ATX power connector and went and plundered an ATX power supply from an old P166. I came back, plugged the plunder into the board’s power connector, connected the cord, and hit the switch. It fired up and the system POSTed. OK, it’s a short in the power supply. I’ll just e-mail Micron with the details and the serial number, and they’ll overnight me another one. In the meantime, this one’s not doing anything anyway.

So I unbolt the bad one, pull it out, flip it over, and get a nice splash of black liquid. What the? 10W40!? In a computer!? Wait… Suddenly the smell made sense. Old coffee. With cream and sugar, judging from how sticky my hands were getting. So I went to the facilities to wash my hands and get some paper towels to clean up the coffee spill that had now migrated to the IDE cables and elsewhere inside the case.

I cleaned up, assembled the system, and e-mailed my boss and my boss’ boss to ask what, if anything, needed to be said or done. My boss is incredibly busy, but my boss’ boss asked if we could loan them another Pentium II until theirs was fixed. I told him he was missing the point: I already got their computer working. My problem with the situation was we had an electrical device with liquid in it and no one told me before I started trying to fix it. The $35 power supply is meaningless. It’s a lot more expensive to repair or replace techs if they electrocute themselves.

He asked me what part of policy isn’t working if it’s not safe to work on equipment.

Am I the only one who remembers from grade school not to put a hair dryer in the bathtub? It’s the same principle, just with more current and less liquid. And I also remember from science class that pure water isn’t a conductor. It’s the stuff dissolved in the water that conducts. St. Louis has hard water. Add coffee. Add cream and sugar. Now you’ve got enough conductivity to short out the power supply. Having some idea what kind of juice accumulates in the power supply (I shook hands with a power supply a few years ago, which is why I don’t open power supplies anymore), this situation strikes me as dangerous.

I was at least owed the courtesy of being told they spilled coffee in there so I knew not to reach in with both hands and complete the circuit. The embarrasment is better than finding a dead Dave laying in their cube next to a dead Micron, isn’t it?

I guess I didn’t explain it well enough.

Hardware mailbag for Sunday

Dave Farquhar Hardware , ,

I wrote up some stuff, forgot to upload it, and lost it. I hate when that happens. So I’ll hit the mailbag for this late Sunday post.
From: Robert Bruce Thompson

Good post on AMD/Intel. At this point, the only thing saving Intel is the fact that AMD doesn’t have enough fab to keep up with demand.

—-

Thanks.

I know AMD and IBM have an agreement dating back to their K6 woes, and I really don’t know why AMD doesn’t have IBM manufacture Athlons and/or Durons to supplement their own capacity. There are issues with someone else making your chips like higher ramp-up time and possibly lower yields, and of course lower profits but I have to think the greater market share they’d gain until they can get another fab built would have to make it worthwhile.

I really wonder if the thing standing in the way of that isn’t technical, but rather Jerry Sanders’ “real men have fabs” attitude.

From: Robert Bruce Thompson

Well, perhaps. But his attitude is correct. Consider CPUs. Intel (bunch of fabs) = dominant; AMD (a couple fabs) = far second; everyone else (no fabs) = non-players. Same thing in chipsets. Intel (bunch of fabs) = dominant; VIA (a fab) = far second; SiS, ALi, etc. (no fabs) = non-players. But I agree that AMD should sub out CPUs to IBM, who can make chips with the best of ’em.

—-

But I wonder if that attitude towards fabs is what’s keeping Sanders and AMD from subcontracting; it seems almost as if using someone else would appear as an admission of weakness and he’s not willing to do it. That’s an Intel-like mistake. When opportunity comes, you have to seize it regardless of how it looks short-term.

From: J.H. Ricketson

Subject: More bargains

Dave –

Another link for surplus/overstock bargains:

www.computersurplusoutlet.com

O have placed one order with them. Completely satisfied, start to finish. Outpost couldn’t have done better. Also, they are in Nevada, which saves me, a Californican, the 8.5% local shopping penalty. They have a good selection of stuff at very good prices.

Regards,

JHR
—
J. H. RICKETSON
[JHR@WarlockLltd.com]

—-

Thanks.

AMD’s turnaround

Dave Farquhar Hardware , , ,

AMD just turned their fourth consecutive profitable quarter, and they say they expect to sell out of Athlons this year. This exposes AMD’s prime weakness: Even though this year Intel has repeatedly failed to execute while AMD has had smooth sailing, Intel has the tremendous advantage of capacity. AMD has two fabs. I don’t remember how many Intel has. Eight?
I was talking with someone before church Wednesday night about new PCs, and he said, “I hear AMD is actually outperforming Intel these days.” AMD, of course, has always been known as the budget chip maker. In 1992, if you wanted a bargain PC, you got an AMD 386DX/40. In the mid 1990s, for a bargain you bought AMD 486s. In the late 90s, you bought AMD K6s. Suddenly, AMD’s not doing much in the low end. They’ve stopped taking orders on K6-2s and they’ll ship their final one this month. The Duron’s a great chip, but they’re not making them in huge quantities. They cite lack of an inexpensive integrated chipset, which is a perfectly valid reason, but there’s another reason. Why should they produce large numbers of Durons? They’re selling every Athlon they can make, so why sacrifice high-margin chips to make lower-margin chips?

It’s been a very interesting year. AMD bet the company on its new fab, knowing that if they made one mistake there was every possibility they were toast. Going into 2000, Intel looked like a company that could do no wrong. But Intel made tons of mistakes this year (the i810 and 1.13 GHz PIII recalls, other high-speed chips that you could read about but not buy, the lack of a suitable replacement for the venerable 440BX chipset, delays on the P4), having a year that made some of AMD’s bad years look good or at least acceptable. Meanwhile, AMD executed. Unlike some, I was fairly confident that AMD would find some way to survive, but survival was about all I expected from them, and about all anyone had any right to expect, given their track record and financial condition.

The guy at church asked if he should sell his Intel stock and get some AMD. I told him I didn’t think so. If AMD can turn around, so can Intel. As for whether AMD is a good investment, I don’t know. They’d look a whole lot better to me if they had a couple more fabs. They’re doing great on CPUs and flash memory, but they need enough capacity to be able to afford to flood the market with Durons, and they need enough capacity to be able to manufacture chipsets if they so choose, rather than developing chipsets as stopgap solutions until VIA and SiS and ALi can step in and then phasing them out. Intel learned that building both chipsets and CPUs allows you to, to a great degree, control your own destiny, not to mention make an extra $20-$30 per computer sold. Intel’s mistake was guessing incorrectly (or perhaps not caring) what consumers wanted, betting on Rambus, and then delivering a bunch of technologies that it was quickly forced to recall.

I don’t know that AMD can turn itself into a giant like Intel, but a strong AMD is good for all of us. It keeps Intel honest.

Hardware developments

Dave Farquhar Hardware , , ,

Hardware news. Lots of stuff today. We’ll take it one at a time.
AMD to hit 1.5 GHz by January. Intel intends to release a 1.5 GHz P4 in late November/early December. AMD’s a bit behind that (assuming Intel will deliver, which they’ve been having difficulty doing lately), but the 1 GHz Athlon performs similarly to a 1.4 GHz P4. Good news for us, bad news for Intel. AMD intends to release a 1.2 GHz Athlon within a month, along with an 800 MHz Duron.

In related news, the AMD 760 chipset (the SMP-enabled one) will be released this year.

The P4 problems are related to the use of PCI graphics cards and Intel has reportedly fixed the problem. Although allowing much higher clock rates, the P4 is less efficient than the P3, so a 1.4 GHz P4 is expected to give comparable performance to a 1 GHz P3. It won’t be until Intel hits 1.5 GHz and higher that the new architecture will give any performance advantage over what’s available now. Not that you can find a 1 GHz P3…

Memory prices are down. If you’re looking to buy, this is a good time. You never know when they’ll rise again, or fall for that matter.

Maxtor buys Quantum. In a consolidation of disk manufacturers, Maxtor bought the disk manufacturing wing of Quantum for $2 billion, making Maxtor the world’s biggest disk manufacturer. Quantum’s tape operations will be spun off into a new company, to be named Quantum.

Mail on modems

Dave Farquhar Hardware

From: al wynn
Subject: External 56K modems

What do you think is the best external 56K modem in the market ?

———-

The best external 56K modems on the market are the 3Com/US Robotics Courier and the Zoom/Hayes Optima series, but they’re extremely pricey. The 3Com/US Robotics Sportsters and the Zoom/Hayes Accuras are nearly as good, cost about half as much, and are much easier to find. I don’t even look at other brands of modems; I stick with the big two (and give the edge to US Robotics).