2007 Volkswagen GTI, Honda Civic Si, and Mazdaspeed 3 Dyno Tests: The Truth in Numbers

In the November 2006 issue of Automobile Magazine, we featured eight cars that point the way toward a new era of frugal fun. Three of them–the Mazdaspeed 3, the Si, and the –are front-wheel drive compacts available right now on dealer lots. All three offer great value, reasonable fuel economy, versatile space efficiency, and a whole lot of power under the hood.

How they produce that power couldn’t be more different. They’ve all got sixteen-valve four-cylinder engines, but the Civic Si makes its 197 hp through spinning its i-VTEC 2.0-liter to outrageous speeds. Volkswagen‘s GTI also displaces 2.0 liters, but it uses a turbocharger to get to the 200-hp mark, making almost 35 percent more torque in the process. Mazda uses the same forced-induction philosophy in the Mazdaspeed 3, blowing enough air into its 2.3-liter mill to make 263 hp.

Three engines that are very different–but which is the most fun? That’s for you to decide. But to help you make that decision, we brought all three of them to a chassis dynamometer. A dyno, as it is commonly called, measures the amount of power that an engine puts to the ground. This number is always lower than the actual amount of power an engine makes, because some of that power is lost to turning the internals of the transmission and differential, the axles, and the hubs, as well as the brakes, wheels, and tires. Knowing the number of horses that actually get to the ground is sometimes more valuable than knowing what’s happening at the engine. After all, what good is a powerful engine if the transmission sucks up half of its output?

Even more important than those peak numbers is the shape of the torque curve. The shape of the curve dictates what you feel when your right foot is buried in the carpet. Some engines make close to their peak torque for a significant portion of their operating range. Others make their peak torque at only one small point–the rest of the time, the engine is delivering only a fraction of that thrust.

Have a look at the individual features below, and see for yourself how differently each of these cars performed. Remember while you’re looking at the charts that the peak numbers tell just a small part of the story–it’s the shape of the curve that dictates how a car feels. If you have any questions, drop us a line in the Automobile Forums and we’ll try to answer your tech inquiries in a way that won’t make your–or our–eyes water.

Volkswagen GTI

Ask any of my friends, and they’ll tell you I’m not a fan of turbocharged engines. I’ve always felt that adding a turbo is a cop-out, cheater way of getting loads of power at the expense of throttle response, because no matter what you read, no turbo gas engine is exempt from lag.

Worst of all, the power delivery and shape of the torque curve is ultimately controlled by the computer. That means that some computer weenie decides how the engine should develop its power. In a perfect world, that decision would instead be made by some mad-scientist engineer in a test car with a bunch of tubes, dials, and protractors. A guy whose neck hair stands at attention when he hears a fantastic, resonating intake honk; whose eyes roll back in his head when his engine has a surge of torque at 4000 rpm; and who laughs maniacally when the cacophony of valves, pistons, and explosions comes to a crescendo at some ludicrously high redline.

Volkswagen, on the other hand, thinks that turbocharging is a great way of getting a lot of power, loads of torque, and great fuel economy out of a small engine. VW has made its fair share of fabulous engines over the years–the old 16-valve and the VR6 come to mind–so their 2.0T merits a closer look.

If you ignore the extremely minor (even to me, the hater of all things turbocharged) turbo lag, you’d never know that the GTI is turbocharged. Except that you’d never get this much torque out of this small of an engine. Floor it at 2000 rpm. Major thrust. 4000? Same deal. 5000? Yup. 6000. Uh-huh. Question it with a stab of your right foot, and the answer is always the same: lots of push-you-back-in-the-seat torque.

Volkswagen rates its 2.0T at 200 horsepower and 207 lb-ft of torque. Our test car put 178 of those horses to the ground, and generated a peak of 193 lb-ft of torque in the process. Those numbers are right where we expected.

Whereas most companies provide peak power ratings at a specific engine speed, VW says that the 2.0T makes its peak horsepower for more than just a single spot on the x-axis, and they’re absolutely right. What that indicates, however, is that the computers took over and castrated the 2.0T. The horsepower curve is suspiciously flat and the torque curve is so perfectly linear that only a computer could be responsible for it.

That’s the bad news. The good news is that-and anyone who has chipped a turbocharged VW engine knows what I’m talking about-there is a whole lot more power to be had out of this engine. Be sure to look for uprated factory 2.0T engines in the future, too. VW might reinforce the engine components, but the actual increase in power will be a result of a reprogrammed computer allowing more boost.

Castrated or not, cheating or not, this engine is immensely satisfying at all times. It sounds great, revs willingly, gets great gas mileage, and makes boatloads of power. It’s a beautifully executed engine that makes good on VW’s heritage of fine motors. You don’t even have to hate the computer weenie.

Honda Civic Si sedan

In our Volkswagen GTI dyno results piece, we wrote that an engine should be tuned by “some mad-scientist . . . whose neck hair stands at attention when he hears a fantastic, resonating intake honk; whose eyes roll back in his head when his engine has a surge of torque at 4000 rpm; and who laughs maniacally when the cacophony of valves, pistons, and explosions comes to a crescendo at some ludicrously high redline.”

Well, guess what? That mad-scientist engineer dude exists. He works part-time at Honda, and he oversaw the development of the Civic Si’s engine. We certainly have a few big gripes about the Civic, especially its electronic throttle mapping, but not a single beef with the Si’s engine itself.

The Si’s 2.0-liter, four-cylinder screamer is rated at 197 horsepower and 139 lb-ft of torque. That’s right at the magical 100-hp-per-liter mark that only the best normally aspirated engines in the world can manage. The trick is, of course, that Honda doesn’t start painting the tachometer red until 8000 rpm.

Horsepower is a function of engine speed–the faster an engine is spinning, the more power it will make for any given amount of torque. Honda’s VTEC ensures that the torque curve doesn’t drop off at high revs by switching to a much more aggressive cam profile at 5800 rpm. The result is peak torque right after the changeover, at 6200 rpm.

On the dyno, the Si put down 180 horsepower at 7800 rpm, right ahead of the VW GTI‘s 2.0-liter turbo four. The torque curve isn’t so much of a curve, it’s more of a flat line with a hiccup when VTEC switches cam profiles at 5800 rpm. As with many of their high-performance engines, it seems that Honda might have been able to make the VTEC cam profile switch at a slightly lower engine speed for a smoother transition. On the other hand, the sudden kick-in-the-pants increase in torque-and sound-is precisely the thing that would make the mad scientist happy. It sure made us smile.

The problem when you’re playing with cheaters (i.e. turbos) is that you often lose by not cheating yourself. The Civic Si swings the same 200-horsepower bat that the GTI 2.0T does, but it isn’t even in the same ballpark where torque is concerned. Mash the throttle in 4th gear at 60 mph in the Civic, and you’ll summon 116 lb-ft of torque. Do the same in the GTI, and you get thrown back in your seat to the tune of 187 lb-ft–that’s 60% more! That said, the Civic Si peaks at 134 lb-ft, and didn’t stray much under 120 lb-ft from 2500 rpm (as low as we were able to obtain reliable data) all the way to redline. Peak torque of 134 lb-ft at the wheels from an engine rated 139 lb-ft is really, really impressive. And arguments about turbo lag here are moot; the Civic Si’s drive-by-wire throttle is so slow to respond that you can literally press the throttle to the floor and release it without any response.

Regardless of its relative disadvantage to the cheaters, the Civic Si has s a truly entertaining engine that’s filled with tech features that (other than the throttle calibration) do nothing to detract from the feeling that someone poured their whole soul into its development. It takes only one trip to the redline to envision that mad scientist thrashing this car mercilessly, laughing all the way.

Mazdaspeed 3

Journalists aren’t exempt from the same childish giggle-fits that afflict regular car enthusiasts. All it takes is a loads of power (preferably enough to overwhelm the tires on command) and even the most jaded journalist will crack a smile. We think the normally aspirated Mazda 3 is a fantastic small car, so we just knew that the turbocharged Mazdaspeed 3 would be a guaranteed giggle.

The Mazdaspeed 3’s engine puts out big peak numbers, and as expected, we loved its acceleration. Many of us noticed, though, that the turbocharged 2.3-liter sometimes felt like an on-off switch. At some speeds, it made amazing power. At others, it didn’t. The dyno chart shows that we weren’t imagining things.

Turbos, by nature, have reduced operating ranges, but the Mazdaspeed 3 seems to be particularly susceptible to that limitation. It has prodigious torque in the midrange, but the engine feels anemic below 3000 rpm or above 5500. Honda‘s normally aspirated Si makes at least 85% of its maximum torque from 2000 to its redline. That means that if you floor the Honda’s loud pedal anywhere between two and eight grand, you’ll have at your disposal at least 85% of the maximum quoted torque. VW’s turbo GTI will hand over at least 70% of its peak number at any time from two grand to its own redline. The Mazda is far behind the others at 20%.

Of course, the ‘Speed 3 makes a much higher peak torque number than either the VW or the Honda (in fact, its peak is almost twice what the Honda puts out). But the GTI’s engine, which is 0.3 liter smaller than the Mazda’s, actually makes more torque than the Mazda from idle to 2700rpm–and again from 5800rpm to redline. The result is that the VW responds to gas pedal inputs with proportional urgency while the Mazda is sometimes fantastically fast, and sometimes–quite surprisingly–not.

A dual-scroll turbocharger like the one in the Saturn Sky Red Line and Pontiac Solstice GXP would help the Mazdaspeed 3 in increasing its usable rev range. As you can see in the comparison graph between the 3 and the Sky Red Line, peak numbers are similar, but the shapes of the torque curves tell two very different stories. GM’s dual-scroll turbo builds boost much earlier, and has almost reached its torque peak by 2000 rpm. At that speed, the Saturn‘s 2.0-liter generates 210 lb-ft, compared to the larger Mazda engine’s 128. Torque is similar throughout the midrange once the Mazda finally wakes up at 3000. By 6000, Mazda’s engine is running out of breath, putting out 138 lb-ft while the smaller GM engine still manages 168.

Punch the accelerator at 3000 rpm in a Mazdaspeed 3 and you’ll probably smile, too. But if you compare it to the more sophisticated engines of the other cars here, you may find it lacking the linearity and broad-range power delivery that set apart the merely powerful engines from the truly magnificent.


Buying Guide

2016 Mazda Mazda3

Fair Market Price $17,990 i Grand Touring (Auto) Hatchback
Motor Trend Rating


29 City / 41 Hwy

Safety (IIHS):

Best Pick

Horse Power:

155 @ 6000