GT750 SLI vs GTX765m for gaming?

What do you guys think would be the best gaming solution for me?

A) GTX765m with Intel Core i7-4700MQ

Or B) GT750 SLI with Core i7-3630QM?

Some people say that GTX765m is the best choice, since SLI doesn't work perfectly in all games, and others say that GT750m SLI would be faster - and better - than the GTX765m, so I'd just like to hear your opinions on which one to choose?

// Spetnazz

 

765m even if SLI does work perfectly. Performance will be about same or better with 765m, lighter, less power, less heat, longer battery life, cheaper.

 

I think we may be in the same situation so I'm hoping for good responses. Originally I was debating between the lenovo y510p ( i7-4702MQ and GT750 SLI) and the MSI GE70-017US (i7-4700MQ and gtx765m). Are you debating between the two as well? Now, after reading a bunch of forums I'm considering the Sager NP8270 (gtx770m (3gb) and i7 4700MQ) for about the same price as the MSI and y510p, my only drawback is hearing bad things about sager. Let me know what you think and what you have heard, heres the link to the sager XOTIC PC - Sager NP8270 (Clevo P170SM) 17.3" Gaming Laptop Another drawback is the MSI has blu-ray and the other two dont.

 

Okay, great. I'll be getting a laptop with 765m then. Thank you so much fo your answer, sir, I really appreciate it!

// Spetnazz

 

I think the sager offers the best bang for your buck, but you sacrifice the portability, backlit keyboard, bluray and better quality speakers of the msi, the sager offers on the other hand a great user upgradeable platform, hands down the better gpu, if you want to add a blu ray it probably costs around 30-40$, if you want to upgrade cpu or gpu down the line you will be able to aswell and you will not void the warranty doing so...
As for build quality I would say sager is on par with msi, maybe even a bit better since the msi is not a GT model .

 

What bad things have you heard about Sager? They are solid machines. They do have their share of minor design flaws, but for the most part are inexpensive and well build powerful notebooks.

You could also consider the NP8230 if you want a 15".

 

And you get the optical bay back for extra storage or w.e you want to do there.

 

750m sli might be a bit more powerful than 765m but I wouldn't go for an SLI build with middle range GPU, especially mobile due to unnecessary heat, higher power usage, bigger power brick and tearing in more demanding games.
With all that being said, the lenovo is more of a bang for the buck kind of purchase, it packs a relatively massive punch in terms of gaming performance for the amount you pay and in some cases you can find discounts which drop the SLI build price below 1k USD which is pretty damn good.

 

750M SLI would be a lot faster for games seeing as it's got the same number of shaders but 300 MHz higher core clock and more than twice the memory bandwidth. Assuming SLI works and scales well, which shouldn't be a problem for most games, then for pure gaming performance, all other considerations aside, the GT 750M SLI wins.

 

The 765m has twice the shaders which is very difficult to make up for in clock speed or SLI even because it's usually only about 60-70% at best improvement with SLI. Where it will improve is with 1080p gaming because each GPU is more or less rendering half the resolution it normally does, so the vRAM bandwidth isn't as much of a bottleneck. But SLI can limit overclocks, and it is yet to be seen how fast the 750m or 765m can overclock.

If you compare 765m 3dmark scores with an actual user score for SLI 750m, you can see difference.

Ice Storm: 765m = 101515 / 750m SLI = 71077 (~ +43% 765m favor)
Cloud Gate: 765m = 17818 / 750m SLI = 12663 (~ +41% 765m favor)
Fire Strike: 765m = 2501 / 750m SLI = 3043 (~ +22% 750m SLI favor)

Fire Strike runs at 1080p, the other two at 720p. I'm thinking the 765m overclock ability will be able to meet exceed 750m SLI even with an overclock.

 

750M SLI has the same number of shaders (384 x 2) along with 200 MHz advantage on the core (1058 MHz vs. 850 MHz) and more than twice the memory bandwidth (2 x 80 GB/s vs. 64 GB/s). SLI scaling is never perfect but it's nowhere as bad as everyone makes it out to be. In a GPU-bound game like Metro 2033 you get almost 100% scaling. Even in a traditionally CPU-limited game like BF3 you still get about 80% scaling. Nvidia's great driver support ensures that SLI scaling will never be a big issue. I'm not familiar with 3DMark 2013 at all but I know Fire Strike is the only test that matters for PC since the other two are cross-platform tests for Android, iOS, etc. In the Fire Strike comparison you referred to 750M SLI is clearly better. I am familiar with 3DMark 11 and the GT 750M SLI in the Lenovo Y510p gets about 5200 GPU score at stock which is about 1000 points higher than stock 765M I think. So even before any overclocking 750M SLI already has an insurmountable lead on 765M, not to mention the latter can't make up the huge deficit in memory bandwidth. If all you care about is gaming and the 750M SLI system is cheaper as well it should really be a no-brainer.

 

You're comparing boost of 750m vs stock speed of the 765m. 750m stock is And Firestrike isn't the only one that matters, lol. What matters in the end are the actual game benchmarks. 3DMark is one of many benchmarks that really tell the whole story. Unfortunately it's the only results we have at the moment for the 750m SLI. And artificial benchmarks aren't usually indicative of actual FPS improvement, especially when it comes to SLI.

And GPU scaling check out this: http://forum.notebookreview.com/sag...eview-wingnut-style-lots-images.html#3dmark06

Compare blue and purple bars. Blue is 680m SLI, purple is 680m single card: http://forum.notebookreview.com/sag...eview-wingnut-style-lots-images.html#3dmark06

Look at the distribution of the improvement from 680m single to 680m SLI, there is an average of 50% improvement, with a large majority less than 60% improvement.





So if you look at single 750m results and multiply by 1.6 it's likely best case scenario for improvement.

 

Check out this overclocked 765m, pretty good score NVIDIA GeForce GTX 765M video card benchmark result - Intel Core i7-4700MQ,Micro-Star International Co., Ltd. MS-16GC score: P4723 3DMarks

Here is overclocked 750 SLI

NVIDIA GeForce GT 750M video card benchmark result - Intel Core i7-3630QM Processor,LENOVO INVALID score: P5396 3DMarks

 

^Both 750Ms overclocked the memory to 1462MHz and the GTX 765M you listed have the memory overclocked to 1150MHz.

Something tells me that GTX 765M isn`t really anything close to it limit while the GT 750M is...
Need better examples than that

 

yeah thats correct the 765 can go higher no doubt about that

 

Since the OP never mentioned what laptops he was considering, if it was something like Y510 at 1080p and the Razer Blade at 900p (33% less pixels). I'd chose the Razer from a pure gaming standpoint. If you were also considering getting work done as well 1080p would help there.

 

Hmmm those results seem awfully suspect. Based on what I have access to, I can tell you I get 80% scaling in BF3, 95% scaling in Metro 2033 and Unigine Heaven, and 90% scaling in 3DMark 11 Basic Edition. Since those results are 680M vs. 680M SLI it's probably CPU-limited. My overclocked 650M SLI is much closer to how 750M SLI performs so it's more accurate.

And the 5200 GPU score in 3DMark 11 is GT 750M SLI stock from the Y510p.

 

I thought the new Clevos had backlit keyboards. Some of the new (cheaper) MSI builds do not come with bluray. Clevo's updatability is also hamstrung by the Clevo's lack of bios support for graphics cards past 1 year. The main advantage of Clevo over MSI seems to be mainly cost.

 

Yes, I am.. Or WAS debating between the Lenovo Y500 - not the y510p since it's not available in Denmark yet - and the MSI GE70, but I've found out that there's a shop here in Denmark that sells Clevo laptops, and since I've hard so much great stuff about Clevo, I might be getting one of those with these specs;

Intel Core i7-4700MQ
Nvidia GeForce GTX765m 2GB GDDR5
Kingston 8GB DDR3-1600 RAM ( Can be upgraded to either 12 or 24 GB)
Disc 1: Samsung 840 250GB SSD (Can be upgraded to a Samsung 840 500 GB SSD)
Disc 2 (Optional - Is not included in the price): Seagate 750GB (SATA3, 7200rpm)

It's a Clevo W370ST, and it's about $180 cheaper than the Y500 and the MSI GE70, so I'll have some money left, which I can use to upgrade to a larger SSD, or to get more RAM if I want to.

 

Suspect? It was using a i7-3720QM CPU, BF3 was running 30 minutes of 64-man multiplayer recording FPS throughout, same map, same server. I just got a Clevo P375SM with i7-4800MQ and SLI 780m. I will do a similar comparison, but in general it's 60-70% max.

I'd like to see actual test results though showing on same system running single 750m, then 750m SLI.

If you want a Clevo with backlit keyboard then you need the laptops that can be equipped with the high end GPU's. The NP8250 that you can get for about $1300 with GTX 770m which costs about same as Y510p and will perform much better than 750m SLI and give you future upgrade to 780m.

 

That's the thing though. 680M SLI/780M SLI is so overpowered for any mobile CPU that you won't get as good SLI scaling as on a much weaker GPU like 750M SLI. 780M SLI is like GTX Titan level performance. And we already know how CPU-limited BF3 multiplayer is. I don't have access to 750M SLI but I've benched my overclocked 650M SLI system in many games and the scaling has been really good. If I have time I'll bench again and make some charts and/or graphs. For BF3 I use an 83-second FRAPS test of the beginning of Operation Swordbreaker from when the doors of the LAV open to when Lt. Cole stops talking at the table. It's impossible to get a consistent benchmark of a multiplayer match and FPS in this singleplayer run is about as indicative of a typical multiplayer match as any I could find.

 

The choice is obvious. 765m all the way.

 

You are right to an extent about the faster GPU's likely won't show scaling as much, but there are many games that are still taxing for even SLI 780m's like BF3 multiplayer, Hitman Absolution, Sleeping Dogs, Witcher 2, Bioshock Infinite, Metro Last Light.

BF3 single player is not even close in taxing as BF3 multiplayer. They're two different games basically. If you play the same server, same number of players, same map, for an extended period (30 minutes or more) you can get a good indicator and check for performance with percentage of frames rendered vs FPS like this:



If Lenovo offered a no questions asked return policy I'd be tempted to order a Y510p with dual 750m and bench the heck out of it and then compare with the 765m I'll be getting in a few weeks.

 

I agree, but it's impossible to find a consistent, repeatable benchmark in multiplayer, short of finding a completely empty server and doing a few laps around the map, and that doesn't accurately reflect real gameplay anyway. Multiplayer matches just vary too wildly depending on so many different factors. I simply don't have the time to spend hours looking for that perfect, repeatable multiplayer scenario and benchmarking it. I chose that section of Operation Swordbreaker because that's about the lowest that singleplayer FPS goes and it's indicative of my average FPS in multiplayer. Without a proper test setup any results are skewed and meaningless anyway. This was the best I came up with that would be as isolated from any biases and irregularities as possible.

 

Running SP is good for SP results. And I do agree it's difficult to get repeatable results with multiplayer, but I've found that doing as I have over the dozens of reviews I've done this is fairly repeatable. It is a good indicator of FPS range and where dips fall and how often. BF3 and MP aside, there are tons of repeatable benchmarks that can be used that relate to the games themselves, and many have built in bencmarks or third party ones that have become standard.

 

Exactly. I chose the singleplayer timedemo because I wanted a benchmark sequence that I had complete control over and that I could replicate every single time. Sort of like a built-in benchmark, if BF3 had one. Multiplayer benches might be useful for finding some general performance numbers and FPS ranges but I wanted something exact. Sure, what I'm doing isn't completely indicative of multiplayer performance, for one MP is a lot more CPU-hungry than SP and minimum FPS is about 10 FPS lower, but it's pretty darn accurate in predicting my average MP FPS. It's great for when I'm overclocking and need to know just how much faster I am over stock.

Since we're on the subject, would it be useful if I provided some benchmark numbers and charts from the Metro 2033 benchmark? That's a built-in game benchmark that hopefully many people have access to. Since I lot of people seem to doubt just how well a mid-range GPU like the 750M can scale in SLI, I'm gonna take my 650M SLI system and overclock it close to or past 750M levels and bench it. If anyone with a GTX 765M system wants to compare numbers that would be very enlightening. Just give me a day or two as I just got a new replacement machine and need to reformat and overclock it.

 

If you have the games, try Metro Last Light, Dirt 3, Grid 2, Hitman Absolution, Tomb Raider, Crysis 2, Sleeping Dogs, Bioshock Infinite, and Resident Evil 6. They all have built in benchmarks, well except Crysis 2 you need the adrenaline benchmark tool, but it's free and works great.

 

I actually have none of those games. I pretty much stick with BF3, Metro 2033, Unigine Heaven, and 3DMark 11 for all my stability and benchmark testing.

 

I have been comparing the gt650m SLI with the gt765m on notebookcheck and it seems that the gt765m gives better gaming performance @ 720p while the gt650m SLI is better @1080p.
And the new gt750m SLI is supposed to be about 25-30% faster that both of these cards, so the choice is really easy here.

 

That's what I've been saying the whole time. Short of getting GTX 675MX or 770M, GT 750M SLI is better than GTX 765M in pure gaming performance.

 

Lenovo does offer a no questions asked return policy.

 

tl ; dr

750 SLI is better in raw perf but 765M works better in games that have bad scaling, and SLI is always prone to MicroStutter.

Go for 765M.

~Owner of a Single-Card Lenovo Y410p

 

You were already given the numbers, in games the 765M pulls ahead.

Notebookcheck benchmarks need to always be taken with a grain of salt.

Also due to input lag and micro stutter SLI needs to have a large FPS advantage to feel the same smoothness.