Book Review: The Anandtech Guide to PC Gaming Hardware

Add Your Comments

What Is The Point Of This Book?

In the author’s words:

The Anandtech Guide to PC Gaming Hardware is your boot camp to becoming a true PC enthusiast and an expert when it comes
to PC gaming hardware. The tools and knowledge provided for you in this book cannot be found anyplace else–they are timeless resources that will provide you
the stepping stone you need to become the guru of the hardware world you’ve always wanted to be. . . .

A major shortcoming of a lot of books is that by the time they hit the store shelves, the majority of the content is obsolete. Although the material in
[this book] isn’t completely immune to becoming obsolete, by focusing on the underlying architecture of today’s technology, you are left with a resource that doesn’t tell you “what,” but it explains “why” . . . .

The bottom line with The Anandtech Guide to PC Gaming Hardware you don’t always learn what is the best out there, but you gain the knowledge to pick the best hardware out there as time goes on.”

Is The Concept of This Book A Good Idea?

We think so, even for those who regularly visit computer hardware websites.

The problem with learning from computer hardware websites is that there’s no real order or structure to it. Learning is a catch-as-catch-can activity, and even a brief perusal of any computer hardware forum will demonstrate that a lot of people
don’t catch much, or catch the important items.

Given enough time, effort and trial-and-error, some people do end up learning quite a bit, but it’s like teaching a child how to read and write by giving him or her a stack of newspapers every day. Newspapers are better at information than basic instruction.

A structured reference book that logically lays out computer basics would greatly shorten the learning curve and save people a lot of time, headache and grief. For many people, that would be well worth $35.

Even those who are very well versed in some or many computer areas could find such a book very useful. Not everyone knows everything about everything; they know what they need to know. If all of a sudden you need to know something you haven’t looked at in the past, a good reference book would save you a good deal of time and effort.

Some will point out that such books quickly become obsolete. While there is considerable truth to this, the Internet offers an easy remedy to that. One can publish a book and provide its readers with addendums and updates on a website very easily. This can lengthen the useful life of such a book considerably. At the least, it prevents instant obsolescence.

“Is This Book For You?”

The author asks this question and says, “Yes.” We say, sadly, no, for anyone: beginner, intermediate or advanced. We’ll explain why in the following pages; it really is a shame.

Email Ed

Why Not?

“Quite possibly the most important chipset release of all time has been NVIDIA’S (sic) first entry into the chipset market with their nForce.” (page 91)

While this rather dubious judgment is easily the most astonishing comment in the book, and a good reason to begin wondering whether buying the book is such a good idea, there are also other, bigger problems with this book lie elsewhere.

A Rush Job

There’s an old saying, “If you want something done badly, that’s just what you’ll get.”

As the author himself says, he put this 500+ page book together in three months.
It shows.

Nor does it look like the publishing company put much time and care into it, either. It’s hard to believe the finished product even went through elementary proofreading, much less any serious editing, and it needed it badly. From poor photography to poor writing, this book gives every sign of also being rushed by Que Publishing.

Many probably suspect that this book is mostly a cut-and-paste job from various Anandtech articles. It certainly looks that way. What most would not suspect is that the whole ends up being much less than the sum of its parts.

All too often, these segments, written at different times for different reasons and in different contexts, are poorly integrated with each other. These different segments thrown in together often give at least the appearance of contradicting each other.

All too often, you have to be rather informed on the subject so that you “don’t read what he says but read what he means.” Otherwise, the comments are unintelligible or easily misinterpreted.

In most cases, it’s simply a matter of a missing fact or short explanation. However, if you know enough to be able to fill in the missing links, you have no need to read the book, and if you don’t, you’re stuck.

Email Ed

To keep this simple, I’ve identified the major problems with this book, and given a few illustrative examples of each. To do more would only make this review much longer and more confusing.


Good technical writing requires clear, precise language. All too often, this book gives you the opposite.

Keep It Simple, Stupid is not the guiding principle of this book.

Phrasing and sentence structure are often awkward and clumsy. Worse, statements are often confusingly or even dangerously vague and sloppy.

For instance, while the book properly points out that shorting out motherboards is a bad idea, saying “motherboard + metal where it doesn’t belong = dead motherboard” doesn’t distinguish between “dead until you get rid of the short” and “broken.”

While a short-circuit can kill a motherboard; it’s not invariably fatal. Indeed, the book goes on as if it had said just that. A panicky person might not quite realize that, and just see that formula. If the book had instead said that a short could (rather than would) ruin a motherboard, that would prevent this problem (along with some useless RMAs and much frustration and fear).

There’s a discussion of how CPU pipelines work and the problems associated with long pipelines. Phrases like “takes a long time” can mean something quite different to the beginner than the nanoseconds it means to the writer.

Grammatical errors are found often enough to be noticeable. You can actually play a little game with yourself while reading the book. Will the book treat the words “integer” and “floating-point” as proper nouns needing capitalization or not? The correct answer is “No, they’re not,” but sometimes the book capitalizes both of them, sometimes it doesn’t, and sometimes it capitalizes one but not the other.

The PIV is recommended in the “High Performance/High Price Monsters,” category. Not a 2.0GHz PIV. Not a Northwood. “Because of its high-speed FSB and memory bus, as well as its hardware prefetch mechanism, the Pentium 4 has the greatest potential to deliver the absolute most in gaming performance.” It goes on to say that’s especially so for 2.0GHz+ PIV, which implies that it’s at least somewhat true for slower PIVs. Maybe that’s not what the book meant to say, but that’s what it said. There’s no mention that AthlonXPs do better at even Quake than most PIVs. Instead, they’re put in the price/performance category.

Too Much, Too Little

All too often, you get told a lot of things you don’t especially need to know, but don’t hear about items that are more important. This happens so often that you wonder if the author(s) know what is and isn’t important.

The biggest omission in this book is the lack of any advice on how to buy components, even after the book tells you to build your own rather than buy from an OEM (advice that is at least somewhat questionable given the ability to order from an OEM and at least largely configure the machine yourself). It’s odd that the website’s own Hot Deals forum, which is one of the best places to find bargains, isn’t even mentioned.

In the CPU chapter, the book has descriptions of “current” CPUs of interest to gamers, but starts off with the Pentium Pro. The book takes two pages to describe it, which is about as much space or more than is given to most of the more modern processors. At the end of this section, the book then says the processor wasn’t adopted by gamers. So why spend two pages talking about it?

On the other hand, the AMD K6 series doesn’t get listed at all and it’s not mentioned even when other processors like the Celeron are discussed. Why?

The cooling section spends over a page discussing the various ways heatsinks are made, but just makes a single passing reference to the high thermal demands of the AMD Thunderbird. It doesn’t discuss at all why high-end CPUs demand high-end cooling. Nor does it mention measurements like C/W which would give the reader an idea which coolers are good and which aren’t.

While the book talks about what happens when video RAM gets pushed too far, it doesn’t mention that too fervent overclocking of memory can cause permanent damage.

These are just a few examples, there are plenty of others.

There is usually little said about future developments in the book. Even when the subject gets mentioned, there are more odd omissions.
Northwood gets discussed, but AMD’s future Thoroughbred isn’t mentioned. Soon-to-come Intel advances like symmetric multithreading are covered, but future AMD plans like silicon-on-insulator are not.

You might think this is evidence of bias, but if you read the book, that’s not the impression you get. There’s no consistent bias; it’s more somebody putting a jigsaw puzzle together that’s missing a few pieces.

You keep thinking as you read the book that they keep almost getting it right, but leave out something. This happens so often you start questioning how well the author(s) really understand (as opposed to know a lot of facts about) the subject.

Email Ed


An instructional reference book should be carefully ordered. Concepts should follow in logical sequence. Too often, this book doesn’t do that.

While the book is structured to start from the beginning, it often brings up terms and concepts out-of-the blue that haven’t been introduced or explained or described.

You see items discussed long before they need to be discussed. For instance, the book starts off by assuming the reader doesn’t even know what a CPU is. For some reason, the author decides to talk about CPU heat by page four. The next thing you know, terms like die shrinks are being tossed around. This is before the poor beginner has any real idea what’s getting hot or how hot it gets or why a die shrink helps. (Even worse, he doesn’t find out anywhere else in the book.)

If a concept comes up which more properly should be discussed in detail elsewhere, the book sometimes makes a reference, then discusses the matter anyway in far more detail than is needed at that point.

Items get placed in the wrong area. The discussion of Northwood is not under the PIV section, but under the Tualatin section.

Infobits get mentioned long after their proper time. When describing the history of the Celeron, the book says that even the cacheless Celeron was popular with gamers. You might think its overclockability might be mentioned as part of the reason, but that only shows up two pages later.

Items sometimes don’t get placed in the proper order. This is not too good when troubleshooting is being discussed. For example, if a machine isn’t working, the first priority is to make sure the one component that if faulty can destroy the CPU or more, the CPU cooler, is working. Not the fourth.

There are a few glaring factual errors. The book says the initial Celerons were made using a .35 micron process. No, it was .25 micron; that’s why they could reach higher speeds than the early .35 micron PII Klameths.

The book compares the Pentium Pro being made with a .35 micron process compared to the Pentium 60 and 66 being made with a .8 micron process. That’s true, but by the time the Pentium Pro came out, regular Pentiums were being made using the same .35 micron process, too.

Not Putting It Together

When it comes to recommendations or even just assessments of current technology, sometimes the book does, sometimes it doesn’t. Even worse, some of the more important chapters that don’t instead provide lopsided descriptions of current technology, which implies recommendation from the sheer weight of words.

For instance, while there is no specific recommendation on chipsets, you do find the comment about the nForce highlighted at the top of page two. That’s followed by five pages of description, which is at least five times longer than the next longest description, and more like ten times the length of the average chipset description. Not a word about the Via KT266A. You want to say it wasn’t around when it went to publishing, sure, but neither was the nForce.

There’s no recommendation on video cards, either. You might suspect nVidia would gather some more halos, but no, ATI gets three times the space nVidia gets.

The author might say that the purpose of the book was not to make recommendations but to teach people how to judge for themselves, but this book doesn’t do that, either.

It certainly mentions items readers should consider (though not necessarily the critical ones). However, when it counts, it doesn’t give readers a way to gauge how important an item is and what kind of impact its inclusion or exclusion has on performance. At least sometimes, you get told about a number of “important” items. How important? Which ones are more important than others? Sometimes, not a clue.

If there’s anybody who has done enough empirical testing to make such judgments, it’s the author and his website. Yet inexplicably, all that data (with just a few exceptions) is never used to give the reader an idea of the impact of various factors.

Even worse, if you go look at the website’s data yourself, you find that the empirical evidence heavily qualifies or even contradicts the assessments made in the book.

For instance, putting the PIV in general as the ideal gaming platform over the AthlonXP looks pretty dubious when you look at the website’s own comparisons.

You’d think the simplest approach would be just to look at the empirical benchmarking data and provide some tips on how it should be interpreted, but that doesn’t happen. Readers get told to go to the website, but they don’t specifically get told to judge the data or how. Rather bizarrely, they instead get told to research unavailable arcane items like “find out how accurate (sic) the BPU [Branch Prediction Unit] can predict branches and compare that to the penalty of a mispredict.” Wouldn’t being given the tools to assess benchmarks intelligently not only be easier, but better?

Email Ed

Not Using Available Resources

Every once in a while, you’ll see a reference to the website for future or more current information, but you might think people who paid for this book deserve a little more than that.

Updates to the book could be included in a different section of the website. At the very least, a page with links to articles of particular interest to those reading particular chapters would be a service, but, no, nothing of the sort.

If the author’s own website and website resources get short shrift in the book, you can imagine how little other Web resources get mentioned. Actually, I found just one reference to another website, at the very end of the book: Adrian Rojak Pot’s BIOS Guide. While the author’s website is pretty comprehensive, there’s at least a few instances where specialist websites would serve as a better resource to readers: Storage Review for hard drives, for instance.

Temporary Silliness?

Finally, there’s the unintentionally funny statement that is just so bizarre and/or shallow that they unravel an already shaky confidence in the book. We’ve already mentioned the “best of all time” nForce comment, but shortly thereafter, you read this:

“The nForce is a two chip solution, but instead of referring to the chips as North/South Bridges, NVIDIA’s (sic) excellent marketing team has come up with two new names: the Integrated Graphics Processor (IGP) and the Media and Communications Processor (MCP).

Hmmmm, Intel came up with new names for its bridges, too, but I guess names like Memory Controller Hub and I/O Controller Hub are signs of a less-than-excellent marketing team. Who cares what they call it?

“A [Front Side] [B]us isn’t something that you can touch or buy . . . . Much like the roads we’re constantly comparing these buses to, you don’t purchase them, but you use them all the time.”

Sure makes what are basically a set of metallic wiring traces on the motherboard pretty mysterious if not downright mystical, don’t you think?

“It is very easy for something as complex as a CPU to overwhelm the uninitiated . . . . Remember how difficult it was to make sure you didn’t break the 10 parts that made up your train set when you were young? Well, now you’re being asked to compare, contrast, purchase, install and tweak a “toy” with 160 million parts.”

Even a little kid has to maintain and at least try not to break something far, far more complex than a CPU. It’s called his body, and even an infant manages at least internal maintenance of that simply because, like a CPU, a body is designed to pretty much take care of itself.

“Hope came for ATI in the form of its Radeon. Not only did the name signify a departure from its previous failures [named Rage], but it also embodied the type of culture that the online gamer was raised in.”

The name apparently is deemed more important than what the card actually could do. If online culture really consisted of shallow marketing using no-names that don’t mean anything pandering to even shallower buyers, we’d all own iMacs. 🙂

Email Ed


For Want Of An Editor, The Book Was Lost

Are the faults I’ve mentioned generally true of the book? Nothing happens consistently in this book. If you don’t like how the book is handling something, don’t worry. It will change.

If writing style were the ocean, you’d get seasick reading this book. Sometimes you go from chapter to chapter and it’s like night and day.

While no large part of this book is completely immune to the problems I’ve described, nonetheless, large parts are more or less OK, and a few parts are actually very good (the description of how video cards work is excellent).

Unfortunately, the whole book should look like that. Even more unfortunately, the poorer parts tend to be the more important parts of the book. Most unfortunately, the worst chapter in this book is the first and arguably most important: the CPU chapter.

Some may think that I’m nitpicking, but the whole point to a reference book like this is to get all (or just about all) the details, get them right, and make them intelligible. Not sometimes nor here but not there, nor even more often that not. Just about all the time.

We are not talking about an item here, a few items there. We are talking about many, many instances of the types of items mentioned above. All together, probably hundreds of instances.

I’ve just given some samples here. If I tried to point out everything, this review would start taking on the proportions of a book, and most of them would in-and-of-themselves be nitpicks.

But it’s the weight of all of them put together that makes this book a quagmire to read and especially to review. It just wears you down, nit by nit.

The real tragedy of this book is that most of these problems could have been fixed with some thorough, competent technical editing and proofreading. They weren’t.

The Brute Force Vs. The Smart Approach

One type of problem, though, isn’t so easily edited or proofread away.

The book promises you to essentially “teach you how to fish” rather than handing you a fish. We think this an excellent idea, but again too often, it does no such thing. Or rather, it teaches a rather odd way to do so.

All too often, there is a curious lack of attention to instructing the reader how to judge what is important and what isn’t, and even when that’s done, the results are often at least questionable.

All too often, this book seem to take the anti-aircraft fire approach to a subject: If you just throw up enough facts at the target, something is bound to hit.

I think most buyers of this book think they paid for a “smart bomb” instead. After all, they can take the anti-aircraft approach for free all by themselves by visiting websites and sorting things out.

This book certainly pumps out facts like AA barrages pump out flak. Not necessarily the ones you need to know, but there sure are a lot of facts in this book.

But facts aren’t intelligence. Facts are just fuel for the engine called intelligence. You can’t go anywhere without fuel, but a fuel dump is useless without an engine. Facts alone don’t matter; it’s how well you use them that does.

On the whole, the simpler the situation and comparison, the more likely the engine in this book will work. For instance, the discussion on monitors, while short, and hardly “explained in greater detail than you’ve ever seen before,” does the job well enough.

Unfortunately, the more complex the situation becomes, and the more trade-offs and factors that need to be considered to make a reasoned judgment, the less likely you’ll get a satisfactory approach or even an approach at all.

If You Build It, Will They Come?

This book is about five hundred pages long. If this book did everything it really needed to do, even after pruning out extraneous or largely irrelevant material and putting in what’s needed, the final product certainly wouldn’t be much shorter than this. It probably would be as long or even a bit longer than it is now.

Whether the book or one like it is good or bad becomes moot if those who need to read it simply won’t, no matter how good it is.

Unfortunately, I fear the real audience for this book hates to read, and will do just about anything to avoid it. There’s a very common attitude out there that scorns book knowledge and thinks experience is the only teacher.

This is bad for this type of book, but worse for the audience.

Trial-and-error is just a stupid, inefficient way to learn and should be avoided whenever possible. How can it possibly be better to “learn” something after five hours of frustrated effort when you could have learned the same thing in much less time just by researching first?

Yes, there are some areas where trial-and-error are unavoidable, and hands-on experience helps. The very act of overclocking a particular set of components requires just that.

But 80% of the overclocking game is decided by what you do before you even order anything, but so many, maybe even most people waste tremendous amounts of time, money and effort because they refuse to be prepared.

There used to be a Fram oil filter commercial showing cars that didn’t get periodic oil filter changes getting towed, with the tower saying, “You can pay me now, or pay me later.”

The most valuable skill in overclocking is not mechanical aptitude, but reading comprehension.

Email Ed

Leave a Reply

Your email address will not be published. Required fields are marked *