AMD Versus Intel Mobile Processor Review

by Dustin Sklavos Reads (226,861)

by Dustin Sklavos

Showdowns don’t get any more hardcore or bitter than this one: Intel vs. AMD. The most popular analogy has been "David vs. Goliath," but I don’t really believe it, although lately it’s been akin to "Godzilla vs. Japan."

Run For Your Lives!!!

I digress. As far as CPUs and platforms go, it’s been AMD and Intel for a long time now. Sure, you can joke about your Cyrix, your VIA or your Transmeta, but the only ones we ever took seriously were AMD and Intel. They’ve had a lot of back and forth over the years, which of course has generally been great for the consumer as actual competition often is, and this past year has been particularly vicious between the two.

Prior to mid-2006 and the release of Intel’s Core 2 Duo on the desktop, Intel had a healthy lead in the mobile market while AMD’s chips were the best desktop performers by far, and unfortunately, AMD didn’t think to try and strangle the life out of its longtime rival by lowering its prices to continue to apply pressure.

Cue 2007, where Intel continues to step on AMD’s head while they’re drowning. Isn’t technology grand?

Intel’s had the best mobile platform for a few years with AMD constantly playing catch-up, blowing their one chance on any kind of lead by delivering the Turion 64 X2 months late and right on schedule to compete with Intel’s mobile Core 2 Duo. Ouch. But Intel’s ruthlessness is our benefit as consumers, because their willingness to reduce processor prices to put the hurt on AMD forces AMD to reduce their own processor prices, and that pretty much brings us to where we are today: two otherwise identically configured notebooks separated by $100 in the marketplace.

So today we’re not going to compete on clock for clock or any of those other BS metrics, because in the end for the vast majority of us, they don’t matter. What we’re interested in is how much we can get for how little we pay. I’ll give you a hint: "a lot."

TESTING

The two competing notebooks are remarkably similar in their configurations and provide a nice apples-to-apples comparison: the AMD based HP Pavilion dv2610us and the Intel based HP Pavilion dv2615us. If you removed the stickers from them, you’d never tell them apart.

PART dv2610us (AMD) dv2615us (Intel)
Processor AMD Turion 64 X2 TL-58 (1.9GHz) Intel Core 2 Duo T5250 (1.5GHz)
Chipset nVidia 7150M/nForce 430 Mobile Intel 965GM Express w/ ICH8-M
Memory 2GB Patriot DDR2-667 2GB Patriot DDR2-667
Graphics nVidia GeForce 7150M (128MB) Intel GMA X3100 (set to max RAM)
Graphic Driver ForceWare 156.65 (HP) Intel 15.7
Sound HD Audio HD Audio
Wireless Broadcom 802.11g & HP Bluetooth Intel 3945ABG & HP Bluetooth
Hard Disk 160GB Hitachi SATA 5400rpm 160GB Hitachi SATA 5400rpm
Operating System Windows Vista Home Premium 32-bit Windows Vista Home Premium 32-bit
Battery 6-cell Li-Ion 6-cell Li-Ion
Typical Price (US) ~$849 ~$899

* Memory was upgraded in each unit from the 1GB DDR2-667 (2x512MB) the notebooks shipped with.

I post "typical price" because the prices of these units tend to vary pretty significantly. For the dv2610us I’m using, for example, I paid $699. But these two notebooks really basically are at price parity, with the Intel one tending to be a little more pricey.

In all testing situations, these two notebooks were configured as identically as possible.

Now, I’m going to get this out of the way right now: I HATE synthetic benchmarks. As I’ve gotten older, I’ve just hated them more and more. PCMark spits out some number that I have a hard time applying practically to anything, and I’ve found that the best purpose these benchmarks serve is being able to check and see if your computer is running like it’s supposed to by comparing with an existing benchmark of similar/identical hardware.

With each test, I’ll explain why I chose that particular program for testing and try to relate it to how you might practically apply it. I’ve also broken down the test suites into three categories: CPU, gaming, and battery.

The CPU tests focus strictly on the raw computational power of the processors themselves, and they fall in line with how a computer might typically be used.

The gaming tests I waffled on a bit, but they make sense to me. The graphics parts themselves may not be that comparable, but as the best and brightest of integrated graphics hardware for their respective platforms, they help to paint the big picture of the kind of performance you can expect from buying AMD or Intel.

The battery test is a simple one: how much battery life can you expect out of this notebook in average use?

CPU TEST 1: WINDOWS MEDIA ENCODER 9 (32-BIT)

Windows Media Encoder 9 is a free program from Microsoft that can easily convert most video files into WMV files, which I’ve found to offer excellent quality in respectable file sizes. Many people prefer DivX and I wouldn’t begrudge them that, but I’ve found Windows Media Encoder 9 preferable for my purposes, and I use it a LOT on the job. This program is one of the two programs I’m using to test the CPU that’s multithreaded; this means it will employ both cores of the processor.

For this test, I took a video file I produced for my job and converted it to a Windows Media Video file. The file itself is being converted with no perceptible quality loss.

SPEC Original File WMV File
Resolution 720×480 640×480
Type AVI WMV
Encoding Microsoft DV Windows Media 9 at DVD quality (VBR 2 pass)
Size 1.15GB 93.6MB

The AMD Turion 64 X2 TL-58 converted the file in 17m:34s.

The Intel Core 2 Duo T5250 converted the file in 17m:31s.

At 400MHz slower, the Core 2 Duo still turns in a pretty impressive performance. The two are apparently so close in performance here that it makes sense they would achieve price parity. In regular use, this would be negligible and unnoticeable.

WINNER: Intel Core 2 Duo T5250. (AMD – 0, Intel – 1)

CPU TEST 2: wPRIME 1.55

The program wPrime is the other multithreaded program in my testing suite, and it tests the raw mathematical computational power of each processor. It’s also a popular program for stress testing processors to make sure they work properly and error free.

It’s more or less the multithreaded successor to the very popular Prime95, so the odd overclocking nerd in the audience will want to look it up when they’re overclocking their computers.

For this test I ran the 32M speed benchmark of wPrime in both single and dual threaded (read: utilizing a single core and then both cores) configurations.

CPU Single Dual
AMD Turion 64 X2 TL-58 81.089 40.889
Intel Core 2 Duo T5250 106.894 53.805

* Time measured in seconds (lower is better.)

As you can see, the increase in performance going from a single core to both cores is effectively double; most programs will never be this efficient, but a program written to make use of more than one core will still see VERY healthy performance gains (typically between 60% to 80%), particularly multimedia software.

What’s notable here is how badly the TL-58 really murders the T5250; that extra 400MHz on the core pays off in spades here.

WINNER: AMD Turion 64 X2 TL-58. (AMD – 1, Intel – 1)

CPU TEST 3: UNREAL TOURNAMENT 2004 (SOFTWARE RENDERING)

A lot of older games can actually run completely independently of the 3D hardware in the computer. While Unreal Tournament 2004 doesn’t explicitly let you do this, you can edit a file to force it to run entirely on the CPU. This proves to be a pretty good measure of a processor’s more well-rounded performance, since in running Unreal Tournament 2004 entirely on it, all of the game’s calculations must be run: physics, AI, sound, and graphics. It’s a nice chunk of "everything."

The game was run at the following settings:
640×480, all settings on Normal, all checkboxes checked, shadows on "blob"

For the benchmark, I used the program UMark, a free program available online that really makes it easy to benchmark the game. The benchmark ran on the Inferno map (one of my personal favorites) with twelve bots to help stress the system.

CPU Avg. FPS Min. FPS Max. FPS
AMD Turion 64 X2 TL-58 32 16 72
Intel Core 2 Duo T5250 37 19 87

It bears mentioning that it wasn’t that long ago that processors didn’t even really have the raw computing power needed to run this game; certainly when the game came out, using the software renderer was a pipe dream and I suspect that’s at least part of the reason it was obscured.

Showing how much Intel’s gaming performance has improved since the Pentium M era, the Core 2 Duo really punishes the Turion here, maintaining both a smoother framerate and a perceptibly faster average despite the 400MHz clock speed deficit.

WINNER: Intel Core 2 Duo T5250. (AMD – 1, Intel – 2)

CPU TEST 4: DBPOWERAMP 12.3 CD RIPPER

This one should be pretty obvious. While most of you use iTunes or Windows Media Player to rip CDs, I’ve elected to use a program I’ve had a lot of good luck with: dbPowerAmp. It’s easy to use, well-rounded, and produces quality MP3s.

For this test, I set dbPowerAmp to rip the MP3s at Variable Bit Rate averaging roughly 160kbps, and set the encoding speed to "Slow (High Quality)." I was originally going to rip the newest CD from The Birthday Massacre, "Walking with Strangers," which I highly encourage people who love eighties music to check out, but that CD is enjoying an extended hiatus in the CD player in my car. I picked an alternative favorite, Toad the Wet Sprocket’s last true album release, "Coil."

Note that the optical drives in the notebooks are specced out identically, though allow for minor variance in the results due to different brands with the same specifications being used. This is actually really common practice for all of the major manufacturers; the drive I got in my dv2615us may very well be the same one you get in your dv2610us, and vice versa.

The AMD Turion 64 X2 TL-58 ripped the entire CD in 5m:21s.

The Intel Core 2 Duo T5250 ripped the entire CD in 3m:44s.

Those numbers pretty much speak for themselves, and the Core 2 Duo’s multimedia performance rears its head here, clocking a full minute and a half faster than the Turion.

WINNER: Intel Core 2 Duo T5250. (AMD – 1, Intel – 3)


GAMING TEST 1: DOOM 3 (OpenGL)

Doom 3 is a favorite of mine, and is going to be my representative for OpenGL games (admittedly a dying breed). It’s got an easy-to-use built in benchmark, and it still stresses modern hardware just enough.

As I’ve mentioned before, the first run of the timedemo in Doom 3 is basically useless since the demo itself caches a lot of stuff on the fly, so I always use the second run after everything’s cached. The game was set at 640×480, Low Quality, with shadows disabled. The game was also patched to 1.3.

The AMD Turion 64 X2 TL-58 and GeForce 7150M recorded a framerate of 36 fps.

The Intel Core 2 Duo T5250 and GMA X3100 with hardware shaders recorded a framerate of 11.2 fps.

The Intel Core 2 Duo T5250 and GMA X3100 with software shaders recorded a framerate of 11.6 fps.

Forgive me for being crass, but nVidia has basically made an ass out of Intel here. While Doom 3 is still pretty punishing on any IGP except nVidia’s, Intel can’t do anything to scrape playability here.

Clearly, the much publicized much improved GMA X3100 is worlds away from what was promised.

WINNER: nVidia GeForce 7150M. (AMD – 2, Intel – 3)


GAMING TEST 2: HALF-LIFE 2 – LOST COAST (DirectX)

In the DirectX 9 corner, we have Half-Life 2: Lost Coast. I actually noodled what to use to represent DirectX for a while, but Half-Life 2 is so stupidly popular, and Lost Coast one of the most system stressing incarnations of it…it just seemed like the right call. It becomes especially relevant since Intel had been really pimping their Half-Life 2 performance back at GDC.

For the benchmark, I used Lost Coast’s video stress test. I set the game to 640×480, all settings on high and with full HDR.

The AMD Turion 64 X2 TL-58 and GeForce 7150M recorded a framerate of 39.68 fps.

The Intel Core 2 Duo T5250 and GMA X3100 with hardware shaders recorded a framerate of 18.72 fps.

The Intel Core 2 Duo T5250 and GMA X3100 with software shaders was completely unplayable, less than 1 fps.

The GMA X3100 is having really nasty growing pains, and that’s evident here. Fog distance seemed a little bit off in parts of the test, and certain effects that didn’t faze the 7150M (like the distorted glass) played havoc on the X3100.

WINNER: nVidia GeForce 7150M. (AMD – 3, Intel – 3)


BATTERY TEST

Believe me, I considered being lazy enough not to include this, but thought better of it. This is one of the major points of contention between the two platforms: which one offers better battery life? So, given similar configurations with the exact same shell, screen, and battery, this seemed like an excellent chance to compare.

To best simulate and maximize the potential battery efficiency of the notebooks, I developed a custom power plan. The screen brightness was set to 20%, wireless networking was enabled, Bluetooth was disabled, and Sleep and Hibernate were both disabled. The screen was also prevented from shutting off. The remainder of the power plan was derived from Vista’s "Balanced" plan. With these settings, I let the notebook "sit there" on the desktop with the screen saver running. This is more or less a best case scenario, but if you’re just using the computer to take notes in class, this is a good indicator of how far you can stretch the battery without periodically putting it to sleep.

It’s also worth mentioning that for the battery tests, I used the original 1GB the notebooks came with, as opposed to the 2GB upgrade. This should have virtually no impact on battery life, as the number of DIMMs in the unit is the same, the capacity is just reduced. If anything, switching to 2GB may improve battery life by a little bit in regular use as the hard drive won’t be hit as frequently.

At these settings, the AMD-based dv2610us lasted 2h:58m before entering sleep mode at 5% battery life.

At these settings, the Intel-based dv2615us lasted 4h:05m before entering sleep mode at 5% battery life.

WINNER: Intel Core 2 Duo T5250. (AMD – 3, Intel – 4)


USER EXPERIENCE

I try very hard to be platform agnostic: I’m not a fanboy for either side, and I’ll use whatever’s the best on the market. That said, I do have certain personal biases. I’ve generally found in my experience that Intel machines – at least, efficient ones – tend to offer "smoother" performance. It’s hard for me to quantify and most people wouldn’t know the difference either way. I also personally prefer Intel chipsets. I’ve found them to be very stable, and their disk controllers have been consistently top notch.

That said, there was a large period of time when nVidia was the de facto standard for performance on the AMD platform. I regret to say that in this case, it feels like nVidia sort of lets down the whole thing. There just isn’t that polish, that "it just works" that I’ve experienced with Intel hardware. Most of this may be due to the depressing performance of the 7150M on Vista’s Aero Glass which I referred to in my review of the dv2610us. Apparently, that was the result of the 7150M’s power management; the clock speed it scales down to on the desktop is 100MHz, and apparently is just not adequate for running Aero Glass smoothly. I’m not sure who to blame for it, but I’m inclined to blame nVidia. Go 6150 based AMD machines don’t share this problem.

Taking the poor Aero Glass performance out of the picture, I’ve found the Intel based machine to feel snappier and more responsive in general. Of course, if you’re not interested in using Vista and want to downgrade to XP, suddenly the AMD unit becomes a lot more attractive. The frugal buyer will likely be able to find the AMD-based dv2610us for around $100 cheaper than its Intel counterpart.

In either case, upgrading the RAM to 2GB should be your top priority. With 2GB of DDR2-667 available in stores for around $50, and on NewEgg for even less, there’s just no excuse not to do it, especially if you’re running Vista. 2GB of RAM is the sweet spot in XP, and in Vista it’s really the minimum to achieve smooth and enjoyable performance. The sweet spot for Vista – at least Vista 32-bit – is probably 3GB, which is impossible in a notebook with only two RAM slots using matched pairs.

Remember that user experience is really key. All the raw horsepower in the world won’t help you if you don’t find using your machine a comfortable, enjoyable experience.


CONCLUSION

I think if you have to pick between the two, the answer is going to be pretty obvious. Even though the AMD chip in the dv2610us is one of the newer ones that draws less power and dissipates less heat, it still has to run 400MHz faster to even think about keeping up with the Core 2 Duo T5250. Intel’s chipsets and processors are, quite honestly, just plain more efficient across the board. At the exact same power settings as the dv2610us, the Intel-based dv2615us lasted more than a full HOUR longer.

Now, that said, the AMD-based unit is cheaper and as far as casual gaming goes, it would get my wholehearted recommendation. The GeForce 7150M is very fast and in practical use I’ve actually found it to be much more desirable than the Radeon X1200 integrated graphics and Intel’s GMA X3100. Remember, you can get an extended battery for HP’s dv2000 and dv6000 series in the store, and if you shop right, you can usually pick it up for $100 and nearly double your battery life.

Unfortunately, I’ve run into a bit of a conundrum as a casual computer and game user, and it has to do specifically with these two notebooks, making it hard to really recommend either one. Here’s why:

For the AMD-based dv2610us, I mentioned in my review of that particular notebook that the Vista Aero Glass performance is inexcusably poor, and at present, there is no reasonable fix for it. What appears to have happened is actually pretty clear: the GeForce 7150M’s clock speed is locked at 100MHz while running in Vista’s desktop, and that just isn’t fast enough for Aero Glass. Yet when you look at the battery life numbers, it’s pretty clear this was about the only way they were going to get decent battery life out of this notebook. In the future, I honestly expect that at some point, a driver will probably surface that will alleviate this problem. That doesn’t do anyone any good in the short term, though, and I’m not interested in buying something because of its "potential." The fact of the matter is that the dv2610us shipped with a real flaw.

For the Intel-based dv2615us, the problem gets a little more complex. First of all, everyone’s been playing the waiting game with Intel’s GMA X3100, hoping Intel would be able to unlock its actual potential. But it’s been ten months since I saw them at GDC, and while I plan to review the part itself in more detail in the future, the fact is that right now, it’s just not consistent enough. Framerates fluctuated wildly, and having to edit my registry to optimize it with certain games is frankly unreasonable. While theoretically you may be able to get better performance out of it in some games right now than other IGPs, and while theoretically it will improve over time, this is all theory. I’m not holding it in my hands. I’m not interested in promises, I’m interested in products.

And then there’s another issue with the dv2615us. Simply put, you can do better. From HP no less! Right now, for $829, you can order the exact same notebook from them if you configure it yourself. This is before using any coupons, and before students apply their (very easy to get) discount. You’ll have to wait a couple weeks for it, but the benefits are obvious: get exactly what you want, take advantage of any deals they offer, and save some green in the process. In fact, for what amounted to about $50 less than I paid in the store, I was able to order the same notebook with a high capacity 6-cell battery, a 2GHz Core 2 Duo, and a dedicated graphics card using a coupon I found on this very site (hint: check the laptop deals button at the top of the screen, it’s awesome).

But we’re not just talking about these two HP notebooks. In the broader sense, we’re talking about AMD vs. Intel, and the situation is grim for AMD. Put simply, Intel’s modern technology murders AMD’s best and brightest. The only saving graces of the AMD platform are their consistently lower prices and consistently superior integrated graphics for the casual gamers. And that said, while AMD’s processors may be "slower" than Intel’s, they are by no means slow. The bar for modern hardware is really pretty high, and the average user probably won’t even notice the difference.

Of course, if you’re not interested in doing any gaming, and you want something that’ll last a while on the battery, suddenly AMD’s value proposition looks a lot less impressive. If it means spending an extra $50 to get an Intel notebook, I’d have to suggest going with that.

And for the casual gamers, it’s important to note that with Dell and HP custom notebook prices so low and constantly seeing coupons and savings, there’s really no reason you can’t get an inexpensive laptop with dedicated gaming hardware. Even the low man on the modern mobile gaming totem pole right now, the nVidia GeForce 8400M G, is light years better than an integrated part. And once you bring that inexpensive dedicated hardware into the equation, suddenly you wonder why anyone would bother with AMD in the first place.

It’s a sad day for AMD, no doubt about it, but remember who benefits from Intel stepping on their head while they’re drowning: you, the consumer. Competition between the two has pushed hardware prices lower and lower, and that’s only better and better for us. With AMD based notebooks hitting $600 and under, it’s incredible you can even get a budget notebook with a dual core processor in it; you shouldn’t even settle for anything else at this point.

AMD vs. Intel? It’s been bloody, and hopefully in a year this situation will have changed. But right now, I think it’s pretty clear who came out on top:

Consumers.


LEAVE A COMMENT

0 Comments

|
All content posted on TechnologyGuide is granted to TechnologyGuide with electronic publishing rights in perpetuity, as all content posted on this site becomes a part of the community.