MSI GTX 1050 Ti Gaming X 4 GB

[HD64G]

A nice review again @W1zzard !

I can see that in other reviews as at http://www.guru3d.com/articles_pages/msi_geforce_gtx_1050_and_1050_ti_gaming_x_review,1.html
there are wider differences in performance between 1050 and 1050Ti (35% more for Ti there) and between Ti and RX470 (40% more for 470 there), so the gamelist is very important now that Vulcan and DX12 have more games on their side.

 

[Jeffredo]

Here they are using a slightly overclocked card. 5-10% extra performance from that alone. Not to mention having external power delivery and better cooling means running at boast clock pretty much at all times vs reference model.

Exactly. The review should have used stock clocks to establish a baseline comparison between a GTX 1050 Ti and GTX 960 4GB. Same with the GTX 1050 review. Once that's done review the factory overclocked custom designs. I've looked over a few reviews that have used stock speeds and that cuts the increase in performance of both cards down to 10% or less compared to the GTX 950 and GTX 960.

 

[W1zzard]

The review should have used stock clocks

Due to how Boost works just setting reference clocks will not result in reference performance. Gotta buy a card that has same base and boost clocks as NVIDIA ref.

 

[hat]

I'd like one (maybe not this particular expensive model) to replace my 660 Ti. The performance and efficiency is much more than my 4 generations old card.

 

[illli]

so.. does not look like much of an upgrade compared to my 660ti, which is on par with the 960. A few frames more, but not much.
Meanwhile I find this review strange b/c nobody in their right mind will try to play games with this card running at 3840x2160. I think it would have been better to do some tests running at a lesser resolution than 1080 instead

 

[BSim500]

Meanwhile I find this review strange b/c nobody in their right mind will try to play games with this card running at 3840x2160. I think it would have been better to do some tests running at a lesser resolution than 1080 instead

Or better still, do all tests at 1080p but test a lot more "mid-weight" games than the usual AAA's (or test AAA's on Med/High). "We only bothered to test only the newest AAA games on Ultra for the sake of consistency" is a pointless excuse for low-end reviews as relevance trumps consistency every time. ie, people who want 60fps Ultra mins in bleeding edge AAA's don't buy bottom rung cards, and people who buy bottom rung cards don't care about 1440p / 4K resolutions, obsess over Ultra presets or even play the newest AAA games. Neither group will care about comparing 1050's vs 1070/1080's or be interested in unplayable results like 18fps 1050Ti vs 10fps RX460 at 4K.

How does Skyrim, Talos Principle, etc, run on these at 1080p? Why do tech sites stop reviewing games that are significantly higher up in Steam's most played list than modern AAA's? Does the Witcher 3's 36fps on Ultra get boosted to nearer 60fps if you drop down to Med/High and if so, how about a couple of screenshots showing any visual differences? Does "Ultra" even massively boost immersion, or is it another Crysis 3 (and many more examples) of being well into the realms of depreciating gains?

^ This stuff is far more useful to budget gamers than simply replicating the same tiny ultra-narrow subset of Doom, DX:MD, BF4, ROTT, etc, out of thousands of widely played games that 19 other sites have done. In fact many budget gamers have long stopped reading such reviews and resorted to amateur Youtube benchmarks showing what sensible settings (with often barely perceptible differences) will give 60fps in practise.

I know we've always had rubbish gaming benchmark "reviews", from testing games at 640x480 (in an era of 1024x768 to 1280x1024 monitors) to exaggerate differences in CPU's to modern day testing low-end stuff at 4K Ultra for the sake of epeen approval (which doesn't even apply to relevant target market), but come on guys, the first site that actually "gets" target market relevancy and uses a little more common sense for low-end hardware will be ad-block white-listed for life.

 

[EdInk]

Are all the titles retested using all the cards or new ones are added to the database? If this is the case, won't there be inconsistencies as driver updates may improve GPU performance over time...just curious.

 

[W1zzard]

Are all the titles retested using all the cards or new ones are added to the database? If this is the case, won't there be inconsistencies as driver updates may improve GPU performance over time...just curious.

• All games and cards were tested with the drivers listed above, no performance results were recycled between test systems. Only this exact system with exactly the same configuration was used.

From the test setup page. Is there anything unclear about the statement? How can it be improved?

 

[EdInk]

• All games and cards were tested with the drivers listed above, no performance results were recycled between test systems. Only this exact system with exactly the same configuration was used.

From the test setup page. Is there anything unclear about the statement? How can it be improved?

Excellent! feel more confident with results. Thanks for clarifying.

 

[ผ่อมด]

GPU overclocking is capped too, at 1911 MHz actual frequency after GPU Boost, but the implementation seems buggy. Even though it shows 1911 MHz, the actual clock can still be increased; at beyond 1911 MHz displayed, the card will run at higher performance, and at even higher clocks, it will get increasingly unstable.

Sounds more like a VBIOS limit and the 1911MHz point is being starved of voltage as it is pushed further and further back along the voltage points of the curve each time the OC offset is increased. Eventually not enough voltage to be stable.

By using fixed voltage (not VF point) so the voltage is constant while GPU clock is free to change we can put in a silly offset and still be rock stable as GPU clock is capped to 1911MHz.

Here is with 1GHz+ offset.

The curve is a little too high to show in AB, just the first 3 points can be seen, however the capped 1911MHz point can be seen. Valley shows the maximum performance clock which is a little over 3GHz and this would be our maximum boost if not capped to 1911MHz. Would also depend on temperature and if able to be stable at that frequency of course.

 

[Nick89]

Great Review as usual.

 

[Clariska]

I got one of these cards as a bday present last month.I personally am truly impressed.I aint much of a gamer reason why i never bothered to upgraded my amd 4170 k cpu lol.But in rise of tomb raider i got 49 fps on max details just with fxaa on 1920x1080 with little overclocking.Thats quite playable.