Rendertest Results - New Laptop

The_Jeff wrote on 11/3/2003, 9:33 PM
I promised a few days ago to post resuls for my new Laptop.

I used the rendertest .veg file from

http://www.sundancemediagroup.com/help/kb/kb_download.asp?id=8


I rendered to MPEG-2 DVD NTSC profile.

On the Dell 1.7GHz Pentium-M - 1 Minute, 54 Seconds
On my Desktop 1.9Ghz Pentium 4 - 3 Minutes and 2 seconds

Both machines have 512 Megs of RAM. In both cases I rendered to the system disk.

The Dell has a 4200 RPM drive (but 80 Gigs so it is generally a newer/fast drive)

These results pretty much match other benchmarks I have seen comparing the Pentium 4 to the Pentium-M

(NOTE: I am talking about the Pentium-M NOT the Pentium 4m or the mobile Pentium 4)



Comments

riredale wrote on 11/3/2003, 9:42 PM
Pretty fast. I think when the rendertest first was posted last year, very few (if any) machines could break 2 minutes.
Liam_Vegas wrote on 11/3/2003, 9:49 PM
It is not the real "rendertest" unless you render to NTSC DV (The_Jeff used the MPEG2 format)

The standard rendertest goes to DV AVI... and I don't think that any 1.7Ghz Laptop can come <that> close to my 3.0Ghz system (which does it in 1m38s).

kdenninger wrote on 11/3/2003, 9:50 PM
What was the video quality set to? "Good" (the default) or "Best"? It makes a significant difference!
The_Jeff wrote on 11/3/2003, 10:50 PM
It was rendered as Good. I'll post updated numbers at best and AVI tomorrow.
The_Jeff wrote on 11/3/2003, 10:53 PM
I'll try DV AVI tomorrow but given that I ran a similar test on a 1.9 Pentium 4 and the 1.7 Pentium-M my guess will be that it does not change things much.

The M has a much shorter instruction pipeline than the pentium 4 and a 1 Meg secondary cache...The relationship I got is not really that far off from general benchmark results I have seen for this processor.
snicholshms wrote on 11/3/2003, 11:41 PM
Does Intel have a an M chip in the 3.0 class?
Liam_Vegas wrote on 11/4/2003, 1:50 AM
Some rendertest resullts on my systems for you to compare with.

The BEST DV AVI format is the one the rendertest is standardized for,

to DV AVI (best)
on my 1.7Ghz laptop (1.5 years old) = 3m27s
on my 3.0Ghz desktop = 1m34s

To MPEG2 DVD (best)
on my 1.7Ghz laptop (1.5 years old) = 3m24s
on my 3.0 GHz desktop = 1m38s

Using the GOOD setting this is what I get.

Rendertest to MPEG2 DVD (good)
on my 1.7Ghz laptop (1.5 years old) = 2m10s
on my 3.0 GHz desktop = 1m04s

to DV AVI (good)
on my 1.7Ghz laptop (1.5 years old) = 2m06s
on my 3.0Ghz desktop = 1m16s

So... certainly that laptop at 1,7Ghz is much quicker than my own older 1.7Ghz... so that processor is definitely quite impressive...
clearvu wrote on 11/4/2003, 7:34 AM
Just had to try the test for myself.

My render result for AVI DV NTSC was 1 minute, 38 seconds for a "best" quality.

I'm running an Intel P4 3.06, 1 gig ram.

By the way, I had a few applications running in the background.
kdenninger wrote on 11/4/2003, 1:08 PM
Thje reason I ask is that with "Good" on my P4 2.4ghz that same test runs in about 1:40, but set to "Best" it takes 2:30 or thereabouts.

BIG difference!

(That wasn't a "quiescent" machine either - so its not "best possible")
The_Jeff wrote on 11/4/2003, 8:07 PM
Ok..here are the updated results (Again 1.7 Ghz Pentium M laptop)

Rendered to DV

1:21 good
1:54 best
Liam_Vegas wrote on 11/4/2003, 9:41 PM
Wow... that is surprisingly close to the render speeds from my 3.0Ghz HT system.

That seems to be an amazing speed difference over regular P4 performance.. so why not introduce this into the regular P4 processor design... wouldn't that also increase the performance of a regular chip? Or doesn't it work that way?

I guess I am trying to figure out what the "catch" is...

I guess my next laptop will be a Pentium-M based chip.
BillyBoy wrote on 11/4/2003, 10:32 PM
I'm highly suspecious of the results. I'm not doubting your testing, rather that the chip is that much faster. If any P4 chip with a clock speed of 1.7- 1.8 was almost as fast as their 3.0 chips Intel would be shouting a lot louder then they are.

What I think you're seeing is it may be faster doing SOME tasks. The test you ran isn't representive of true rendering and was mainly used to stress Vegas. I would suggest you next try it rendering a real world project then render again on your older P4 desktop and see if the performance holds up.

Anyhow, for those that want more:

http://www.intel.com/products/notebook/processors/pentium4-m/index.htm
Liam_Vegas wrote on 11/4/2003, 11:08 PM
In one way I would like that to be the case... after all... why buy a 3.0Ghz CPU when a 1.7Ghz M series is that good.

But I think the rendertest.veg seemed to be a pretty good guage of general rendering speeds right and it seemed (up until this one test) to be a good indicator overall through the range of CPU speed. So.. why does it seem so much more proportionally faster on this chip?

I am also a little surprised at the results that "The_Jeff" has reported hence why I am asking what is really going on here. Because so far it seems somehow this chip is breaking the laws of physics (exagerrating a bit there I know.. but you probably get my meaning).
The_Jeff wrote on 11/5/2003, 10:09 PM
It certainly may not always be as close a race as rendertest would have one believe. But it is not that out of bed with other reviews I have seen.

http://www.tomshardware.com/mobile/20030205/centrino-13.html

Shows a 1.6 Pentium-M being faster than a 2.2 Ghz Pentium 4 so the fact that a 1.7 Ghz Pentium-M is not that far behind a 3.0 Ghz P4 is not that surprising.

As for violating physics...not really.. Double the secondary cache and and shorted the instruction pipeline and one can do some interesting things..It is one of the reasons (for example) why the PowerPC G4/G5s either come close to holding their own or beast higher clocked P4s (that and Altivec)..In any case..please don't let this degenerate into a PowerPC v.s. Pentium thing..


As for why one would not just buy a P-M...They are expensive...And they top out at 1.8 Ghz at the moment (one of the things that lets the P4 run so fast is the long instruction pipeline) so it is unlikely that a -M processor will ever run at the same max speed as a P-4.

So, I guess the only place I am headed with this is if you are looking at a laptop, consider a -M processor in the mix so you don't end up having to live with horrible battery life for a small % increase in performance..