Equivalent: 1.9GHZ AMD Turion Dual Core?

[Jerry Jones]

What is a 1.9GHz AMD Turion dual core equivalent to on the Intel side of the equation?

Anybody know of some kind of chart or table that shows how it compares?

Thanks,

Jerry Jones

http://www.jonesgroup.net

[Hulk]

Clock-for-clock in encoding applications and other video related tasks the Core 2 Duo is about 20% faster or better. Of course in applications that rely more on hard drive performance and other system bottle necks the delta will be less. Those metrics usually involve business oriented tasks but most people aren't waiting on MS Word or Excel.

I would say a 1.9 Turion would be equal to a 1.4 or 1.5GHz Core 2 Duo. You just can't beat the C2D for performance or battery life since you can get a slower clocked C2D that will still perform better than a higher clocked Turion and use less power.

It's not that the Turion isn't a good chip. It is. The C2D is a great one. Especially in mobile form. Remember the Core Duo was initially designed by the Intel Israeli team specifically for low heat/low power applications. The C2D improves on that design.

Have a look here: http://www.anandtech.com/mobile/showdoc.aspx?i=2856&p=7

[Marshmallowman]

The other thing I have noticed when scoping out laptops, is most turion solutions seem limited to 2G of memmory, where as C2's usualy have 4G max memory sizes.

[Jerry Jones]

OK, then how would a 1.4 or 1.5ghz Core 2 Duo compare to a Pentium 4 3.4ghz?

Jerry Jones

http://www.jonesgroup.net

[Marshmallowman]

something like this(red ones)

http://www23.tomshardware.com/cpu_2007.html?modelx=33&model1=909&model2=884&chart=431

[Jerry Jones]

Thanks.

I wish the manufacturers would develop a better system of advising the consumer.

It's way too confusing for the consumer to make an intelligent choice.

I'm tired of it.

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

Tom's also shows the comparison on the "mobile" chart here:

http://www23.tomshardware.com/cpu_mobile.html?model1=704&model2=703&chart=267

It seems to me the AMD Turion actually is better compared to the older Intel "Core Duo" and not so much to the "Core 2 Duo."

Jerry Jones

http://www.jonesgroup.net

[Hulk]

Tom's also shows the comparison on the "mobile" chart here:

http://www23.tomshardware.com/cpu_mobile.html?model1=704&model2=703&chart=267

It seems to me the AMD Turion actually is better compared to the older Intel "Core Duo" and not so much to the "Core 2 Duo."

Jerry Jones
http://www.jonesgroup.net

The Turion is not better than the older Core Duo. At best the Turion can match it. And the Core 2 Duo is a significantly better (wider) architecture. Like I said the Turion is a fine chip but the Core 2 Duo man handles it in just about every way.

Keep in mind that Merom and Conroe, which are both C2D, are essentially the same chip with the mobile parts (Merom) having additional power saving features and generally being cherry picked chips that run on low voltages, which is obviously important for mobile applications.

Turion are basically the same thing but are Athlon X2 parts.

We all know how the Athlon X2 compares to the C2D, it's the same with Turion and Merom.

As for your question about the P4. A C2D in processor related tasks like video encoding is just about twice as fast as a P4.

So, 3GHz P4 = 1.5GHz Core 2 Duo = 1.9~2.0GHz Turion/Athlon X2

The new naming scheme is confusing. Once IPC's (instruction per cycle) weren't equal AMD had to go with the new naming scheme to try and sell their lower frequency yet faster Athlons against the P4's. So they pitted the Athlon 2600 against a P4 2.6, etc...

Then Intel ran into thermal problems with the P4 at 3.8GHz. So the chip that was supposed to scale to nearly 10GHz topped out there. Meanwhile the Athlon continued to scale happily to 2.8GHz (and a bit beyond). The 1 GHz delta between the Athlon and fastest P4 was not nearly big enough to keep Intel in the lead and the Athlon became the performance champion over the P4 a few years ago.

Intel was screwed. They touted megahertz is king for so many years and now they were thermally limited with the P4. It was at a dead end and AMD was killing them. Keep in mind that when I say thermally limited the P4's transistors could open and close fast enough to scale very high, but the heat was too much. A few years ago Intel actually demonstrated a P4 running at some ridiculous frequency (I think like 10GHz), but that chip was liquid nitrogen cooled. You see they had an imbalance in the design. The deep pipelines could scale to very high frequencies but the heat was too much and heat kills a chip.

All the while Intel had the Irael design team that had been working on the Yonah chip. Otherwise known as Core Duo. Yonah featured a much shorter pipeline than P4, (12 stage vs 30 for Northwood and I think 41 for later iterations of P4 like Prescott), and was "wider" in that it could process more instructions simultaneously. This chip was already comparable per cycle with Athlon X2 so they moved development to that architecture.

They ended up making Yonah even wider and just a bit deeper (14 stages) and added a bunch of other really neat architectural improvements and ended up with Conroe. What we now know as Core 2 Duo.

But of course Intel had a marketing problem since they could no longer market by MHz. Now they were faster per clock cycle than AMD!!! The tables had turned! So they created new naming conventions that have no comparision to AMD chips, or frequency, only to other chips in the same line. For example, E6600 is faster than E6400. Both are Core 2 Duo, one at 2400 the other at 2133MHz, respectively. "E's" mean dual core, "Q" means quad core.

Now to be totally honest about this the "2600" or whatever number in the Athlon XP line wasn't officially supposed to be compared to a P4. It was supposed to compare performance to older Athlon chips. But that was just a line I think.

Now there may be errors in this so please don't come down on me. I wrote this from memory. I believe most of it is correct if somewhat watered down.

That's how we got to this naming convention.

When Barcelona launches in the very near future we'll have yet another naming convention and possibly a chip with a higher IPC than Conroe.

But then Penryn will launch on Nov 10 (I believe) and will be about 5% better IPC-wise than Conroe. With new SSE instructions in video encoding it will really be amazing as early benches have shown. Penryn being the first 45nm process chips will also be cooler, faster, and use less energy than Conroe.

Intel is very smart. They build a new architecture on a proven process technology. Refine the architecture on a smaller process and then go to a new architecture on that process. For example.

The last P4's were built on a 65nm process. By the end of the P4 run Intel had the bugs worked out of the 65nm process and were getting excellent chip yields.

The first Conroes were produced on this 65nm process, thereby eliminating one unknown, the process technology from production of Conroe. As we saw Conroe came right "out of the oven" with good yields, as indicated by chips overclocking 50%.

Now with the move to 45nm Intel will evolve Conroe to Penryn. Not an overhaul but a lot of small improvements. The big change is the move to 45nm.

Then in a year they will do a new architecture, Nehalam, with big changes like native quad core and integrated memory controller among other things. But they will start that chip at 45nm and evolve it to 32nm.

One of the problems with Barcelona I suspect is that AMD only just moved to 65nm and they are doing Barcelona on 65nm so they have compounded their problems by doing a new architecture AND a new process.

Have you noticed that the AMD 90nm parts are clocked higher than the 65nm parts? This is because AMD had things more under control with the 90nm process. Sure they're 65nm parts are cooler and more efficient but they still don't clock as high as the 90nm ones. But the 65nm parts are doing better every day.

I'm exhausted.

[Jerry Jones]

Thanks, Mark.

You know, as I contemplate computer upgrades, I'm struck by something.

The system requirements are going up... not by just a little bit... but by a large margin.

If I were to upgrade to a new Windows-based system (VISTA), I would think that the 64-bit version would be best for me.

But, if a person chooses the 64-bit version of VISTA, then that almost guarantees the need for 4GB of RAM.

And a discount processor won't cut it either.

On the Macintosh, I'm thinking a person can do very well with just 2 or 3GB of memory.

These cheaper Windows-based PCs look pretty good until you realize the slower AMD processors are probably the reason for the steep discount.

And the Windows operating system version usually isn't the 64-bit version.

Jerry Jones

http://www.jonesgroup.net

[Jerry Jones]

On the other hand, this guys compares Windows VISTA 64-bit to "Windows ME."

Here:

http://www.notebookreview.com/default.asp?newsID=3529

Jerry Jones

http://www.jonesgroup.net

[Hulk]

Jerry,

Here's my take on it. I'm staying with XP until I move to Vista 64. And I'm not moving to Vista 64 until all of my apps are 64 bit. Or at least most of them. I'd like to see a reason to go 64 bits and that reason is performance.

As for the processors the Athlon X2 processors are pretty darn good and the prices are great now. You just have to keep in mind that a 500MHz or 600MHz slower C2D will perform as well.

Core 2 Duo's are also good deals right now. For $280 you can get a 3.0GHz dually.

But the deal of the century for video guys is the Q6600 which can be had for $265 right now. That 2.4GHz of Quad core power. Plus simply raising the FSB from 266 to 333 will put you right at 3.0GHz with just a touch more Vcore. Add in a good cooler and you have a $1000 processor. 3.0GHz is a relatively conservative overclock for the processor I might add. 3.2 would be pretty easy to attain. Over 3.2GHz is when you start pushing the Vcore and your cooling solution.

For around $200 you can get a very good performing Core 2 Duo and you don't have to overclock it.

I'm going to be upgrading to the Q6600 and taking it up to 3.2 one of these days. That will effectively double my computing power for applications that can utilize all 4 cores.

In a year or two well be looking at 8 core Nehalem parts that will have better IPC than current C2D's. If you are running a P4 right now imagine 16 times the computing power. My current C2D benches 4 times faster in Vegas than my old P4. The quad would be twice as fast again. And 8 core Nehalem over twice as fast again. Right now I can edit AVCHD on my system with good performance. In a year it won't even be an issue.

[Hulk]

I forgot to mention that DDR2 memory is so inexpensive now that it's not a big financial concern.

I just picked up 2GB of PC6400 DDR2 for $50.

[Hulk]

Jerry,

Your question got me to wondering how my new system would compare to my old 3.06P4 running my old MSPro 7 test.

Well the numbers are pretty amazing.

Compressing the old test project to MPEG-2 times as follows. Keep in mind there was lots of effects, PIP, text, moving pictures, etc in the test file and it worked with DV files.

P4 time: 612 seconds (7.3 fps)
C2D time: 253 seconds (18.6 fps)

Now I realize that my C2D is at 3.2GHz and it is dual core but you have to keep in mind that the CPU usage never went above 56% and for the most part was right at 50%.

Clock-for-Clock my new system is at least twice as fast as my old P4.

I also used to computer an efficiency number for this test. It was the number of clock cycles needed to render the test file at 1fps. Of course the lower the number the better since that meant the processor was doing more work per clock cycle.

Here are some average numbers from the systems reported back in those days.

P4 - 320
Athlon XP - 260
C2D - 172

My P4 used the slower 533FSB and was there for less efficient than the 800MHz FSB P4's but still the Core 2 Duo is nearly twice as fast as the fastest P4's at the same clockspeed. Also keep in mind that the Athlon X2 of today is basically the same chip as the XP except for dual cores.

I'm telling you the C2D is a monster.

It's funny going back in time like this. It told my old system 10:12 to render a project my new system renders in 4:12 and it's doing it on one core!

Of course preview is real time even through the most difficult sections of the project even with the "multithreading preview" set to 0.

Of course you can always bog down a system if you really try but for all intents and purposes editing DV with a Core 2 Duo at 3GHz is a realtime experience even with pretty heavy FX applied. Just about the only effect that will choke this system is previewing an unsharp mask filter. That one is a real bear.

[wkulecz]

It's way too confusing for the consumer to make an intelligent choice.

I'm tired of it.

That's the point, they just want to wear you down so you spend as much of your money as you possibly can. If they came out and said the extra $100 for DDR3 gets you 5% better performance than DDR2 I doubt anyone spending their own money would go for it!

--wally.

[Hulk]

Wally,

It's true you do have to be kind of up on things to make sure you don't buy technology that has no real performance benefit.

DDR3 in my opinion at this point in time is a waste. DDR2 is much cheaper and basically performs as well.

P35 Mobo with a Core 2 Duo in the $220 range with 2GB of DDR2 is a fine system perhaps just a bit off state of the art, but costing about 1/4 of the price.