650M Preliminary Benchmarks = up to 67% faster than 555M

Been playing with my new R2 for the last day or so and wanted to share some quick benchmarks I ran. Important to note the graphics card did not exceed 60 degrees throughout benchmarking!

System:
3720QM
512GB Samsung PM 830 SSD
Nvidia 2GB 650m
1600x900p screen
8GB DDR3 1600

EDIT: Thank you to R1 users who have added to this thread with their R1 OC results! I have added your contributions to each section. I'm all for accurate results and running the 555M tests with the latest software. To be fair though comparing the 555M with launch drivers seems fair to compare to the 650M with launch drivers since it is just as likely the 650M performance will improve as their drivers also mature! Apples to apples OC to OC the gap is down in the 35% range. I have edited topic title to reflect.


The 67% # comes from my comparison to Anandtech's review of the M14xR1 specifically Metro 2033 since it is perhaps the most taxing benchmark I have:

Base 650M with no overclocking is 46% faster:



Overclocked 650M to 880 Core and 1203 Memory is 67% faster:





Thank you member Lancer for posting 555M OC results: http://forum.notebookreview.com/8501016-post23.html
555M OC vs 650M OC gap closes to 35%
His results account for updated drivers and I imagine some quality OCing time. I am using MSI Afterburner that is limiting my core clocks to 880. The usual tweaks to lift restrictions in software are not working, but due to the stability of the 880 I think there is much more headroom!

 

Finally some real 650M GDDR5 benchmarking, thanx +rep

 

Great to finally see some benchmarks. The r2 does seem like a good improvement.

 

3d Mark 11 - P2299 non OC
NVIDIA GeForce GT 650M video card benchmark result - Intel Core i7-3720QM Processor,Alienware M14xR2 score: P2299 3DMarks

3D Mark 11 - P2624 with OC
http://3dmark.com/3dm11/3360467;jsessionid=11bg5g8wswg221pzzk1a4qkegx

Thank you member shnarf for providing 555M OC results: 835/1080 - P2002
http://forum.notebookreview.com/8501017-post24.html

555M Put up a P1315. Non OC 650M is 75% faster and OC 650M is 99% faster.

 

Was the 555m at the lower resolution? Not sure what difference that would make. Still a pretty nice jump though!

Any chance of a quick mention of battery life? The guy who did the unboxing on YouTube is claiming 10 hrs browsing and up to 5 gaming! Hard to believe so it would be good to hear what others are seeing

 

All benchmarks were like for like. 3DMark11 has a built in benchmark test that is what everyone uses to make sure the scores are standard and fair. For Metro 2033 I matched the settings used in the Anandtech review as seen on their screenshot and mine. Just tried to do a Stalker Call of Pripyat benchmark, but there are too many setting for me to know exactly what was used in the M14x 555M benchmark so I won't be using the results.

As for battery you can see in my screenshot my battery came at 10% wear level, so I am already starting at 90% new capacity out of the box! I will contact Alienware about that one. As for mobile gaming I have not had an opportunity to really test out that battery.

I will say this is a huge improvement in lap usable gaming as opposed to my M18x!

 

im curious does it still downclock the gpu like crazy on battery with the r2?

 

wow after these results my love for my R1 started to disappear

 

ofcourse it does it to increase battery life when in battery mode.

 

You've compared 650M (not overclocked) to 555M (not overclocked) and 650M (overclocked) to 555M (presumably still not overclocked); any idea how it compares vs an overclocked 555M? like 650M OC vs 555M OC or 650M vs 555M OC? Would be useful info since a lot of people like to OC the 555M by a good bit!

 

Thanks a lot for the benchmarks! +rep
I can't wait to receive mine, I'd like to see if the scores are the same with the 3610QM, 6GB of ram and a normal HDD (my specs)

 

Correct on the non-overclocked 555M, I do not have one to add the results. I can only offer an apples to apples non-oc results from the M14x reviews online and show people what the 650M can really do. Would love for someone to add their 555M overclocked results!

 

The 650m was running at 60°C max OC or non-OC?

 

yes but shouldnt the 650 be more efficient? so that even when it downclocks, it wont cripple it like the 555m

 

Non OC. After 1 hour of Metro 2033 overclocked (100% stable by the way) max temp was 65. My laptop is elevated on a cooling stand with fans directly blowing into the intake so I expect it will be hotter on a lap with air intake slightly obstructed.

Not all that familiar with the 555M downclocking issue. What did you have to do to trigger this behaviour and I can give it a try later.

 

I'll add my oc 555m results in a moment

 

it just downclocks it on battery and cripples it pretty badly.

 

Yeah when on battery it runs game like if they run on intergrated graphics :/
EDIT: To give you an idea look at this thread: http://forum.notebookreview.com/ali...ne-run-3dmark06-battery-555m.html#post7691272
12566 with 3Dmark06 plugged in with the 555m
4036 with 3Dmark06 on battery with the 555m
2843 with 3Dmark06 on battery with the integrated graphic card.

 

It's good to see such results! Thanks for the comparison .

When you have time if you could run Bf3 and Crysis 2, possibly to see what settings you can run the game at max.

Hopefully with the 650M if I end up getting the R2 in September I can max out Mw3 without overclocking GPU ( I had to run 2x AA @ 720p and an overclock of atleast 735 MHZ to get 60-90 fps with dips to low 50s around smoke).

 

LIttlezip - NM... found the tessellation and DF settings in your screenshot. Thanks!

will post 555oc figures here shortly. The anandtech figures are old based on old drivers. We all know that drivers make a huge difference. So I will post stock vs oc (@750/1080) standard gaming clocks.

 

Yeah but about the drivers it will probably be the same for the 650m, we'll see for sure improvement in the future with new drivers.

 

You will probably get something around 1800.

 

http://s7.postimage.org/6lujoqdxn/Metro_2033_oc750.jpg

oc vs 650 stock is 17 percent higher for 650. Score was actually 30.94 (avg 3 runs) but I will round down as the author did. So 30 x 1.17 = 35.1 (17 percent).

oc vs 650 oc is 35 percent. (30 x 1.35 = 40.5) cut the 555 a .1 slack here.

All in all the 650 is a definite improvement over the previous 555. Whether or not its a 400-500 dollar upgrade is up to you. Assuming you sell your 555 for 700-900 and then purchase a new R2 for a decent rate. If you are a new alienware owner the m14xR2 looks to be a great buy. Congrats to the new R2 owners!

 

Best I have been able to put up with the 555 in my R1 is P2002 oc 835core/1080vram. I am happy to see my R2 will bump that up another 250+ even at stock speeds. I was really thinking it would suck if the huge overclock on my R1 translated to the same performance level.

 

I have to give the 555M some respect, closing the gap when oced. I honestly never saw a laptop video card overclock as well as this card did I had it stable 770 mhz without voltage!

Either way hopefully we see the 650M take on 200 MHZ oc soon Maybe with volt mods or just random cards (I know some oc better than others, when I replaced my motherboard I had to replace the 555M as well and lost 50 MHZ stability ).

Anyways if it's a 200-300$ upgrade fee, yeah I guess its worth it, anything more...meh!

Lancer said it best, new comers are in for a treat

 

Whats a safe overclock for the 650m and what program should I use?

 

Why does the GDDR3 version of 650M have a better score?
You would think the GDDR5 650M would have scored much higher.
You can thank HTWingnut for being kind enough to run some comparo benches. All the credit goes to him.

 

Second request, since I'm one of the few who OC to the 800/1600 point, just want to see how much improvement between the 650M vs 555M OC.

For the first question, I cant answer, for the second, there are 3 well known -> MSI Afterburner, EVGA Precision and NVIDIA System Tools. Myself I use the NVIDIA System Tools but it is not so different with the other 2, just that you dont have to have a separated program since the NVST will be integrated with the existed NVCP.

 

I don't think 3Dmark is a good benchmark...I've never trusted synthetics to show accurate scores. I think ppl should take things witha grain of salt when it comes to 3Dmark.

@Ikas_V

I think it's because of the Higher core clock sported by the DDR3

 

Yup, the GDDR5 version is clocked at 735mhz core while the GDDR3 version is clocked at 850mhz.

 

I ran mine and only got P2322 3DMarks

 

Some Heaven results:

Stock clocks:

Spoiler

OC:

 

OC'd using the msi after burner GPU 880/memory 1200 and I got 2707 on 3dmark11

 

Not to hijack the thread. But in case you guys wanted to see how the 650m compared to a 560m in heaven scores, tesellation is off for these. 650m is a pretty sweet deal in a smaller package.

This is a stock 560m at 900p resolution



Stock 560m at 1080p



Oc 560m at 1080p

 

Impressive, it seems maybe the 560m is still stronger since it can semi keep up with resolution?

 

I'll run another bench later tonight with tesellation on to see how big of an impact it has. I wonder if the 650m could see an added performance boost once the drivers mature a bit.

 

Hi, do any of you guys know how if hyperthreading comes pre-enabled on the M14xR2 laptops, and if yes, is there any option in the BIOS to disable it? I read somewhere on this forum that one could boost gaming performance while keeping heat at a stable level by disabling hyperthreading and enabling the intel turbo boost

 

Just ran Unigine and got the following benchmarks

First one is stock and second is overclocked (+50 MHz with msi afterburner)




Let me know what you think, again they are just numbers to me =)

 

Why is willz so much higher than littlezip? Quality benchmarks = full disclosure.
Driver version
Any OS optimization and settings
GPU clock speeds (screenshot)
CPU clock speeds (screenshot)
Cooling (stock, pads or other)
Any other details that are relevant to the benchmark

 

how does the 650m compare to the 660m?

 

The 660m is about 15% faster. The 660m and 650m gddr5 differ in clocks only.

 

would the 1GB DDR5 650m differ much from the 2GB benchmarks we see here?

(slightly edited)

 

The DDR5 version of this card tends to generally be faster than the DD3 version, regardless of the total amount of available memory. Graphic Double Data Rate ver. 5 (GDDR5) is an improvement over its DDR 3 counterpart and comes at a higher non-OC memory clock rate of 1000Mhz, one hundred Mhz more than the DDR3 version.

 

Man, that 650m makes my old 460m look slow now.

Edit: The 3dmark 11 score in my sig is with my 460m, haven't got around to pushing my 6990m with 11 yet. Been to busy

 

oh I'm so sorry, I meant to type DDR5 originally :s
Regardless, thanks for the reply, I was wondering how much of an improvement was from the DDR3 anyways
The m14x r2 just becomes more desirable for me day-by-day!

 

+1 to this, its got the performance of a last gen m17x in a lot smaller package

 

Absolutely! the M14x R2 is a beast I'm glad I could help ya out!

 

So what would be a safe clock for the r2? I have the 3610 IB processor, standard 7200 rpm HD, and the 2gb gt 650m on my config, but I don't have any extra cooling accessories, just OC diamond paste on the gpu/cpu and the internal stock cooling system. Would I be able to safely overclock to 880 mhz and 1203 memory as littlezip did?

 

Just slowly up the clock speeds till either the temperature gets to high or the system gets unstable. Every card overclocks different so it will take some experimentation.