[swaaye]

I imagine there's still a NV preference out there, which AMD has been solidifying lately, and that price is the other primary consideration. I don't think very many gamer people care to or have the capacity to consider other factors even if there are reviews to try to comprehend.

If there is big feature disparity, like when AMD had DX11 and NV did not, I'd say that can change things.

[madyasiwi]

Originally Posted by swaaye

I imagine there's still a NV preference out there, which AMD has been solidifying lately, and that price is the other primary consideration. I don't think very many gamer people care to or have the capacity to consider other factors even if there are reviews to try to comprehend.

You really can't deduce that from the last piece of information alone. If Newegg sold 500 unit EVGA 550 Ti's, 250 unit Shappire 5770s and 350 unit Shappire 6770s; the EVGA one would still be the top selling card despite the fact that a 6770 is nothing but rebranded 5770

[Malo]

Would that make the G92 the best selling card ever?

[trinibwoy]

March results are out. Interesting development:

There was a bug introduced into Steam's survey code several months ago that caused a bias toward older systems. Specifically, only systems that had run the survey prior to the introduction of the bug would be asked to run the survey again. This caused brand new systems to never run the survey. In March 2012, we caught the bug, causing the survey to be run on a large number of new computers, thus giving us a more accurate survey and causing some of the numbers to vary more than they normally would month-to-month.

Post the correction there were some significant swings:

Big jumps for

* Windows 7 64-bit
* Intel graphics
* DX11 graphics
* AMD CPUs

Big drops for

* The English language
* Quad-core CPUs
* nVidia graphics
* Evergreen (5770, 5850, 5870)

[Davros]

* AMD CPUs - suprising

* Quad-core CPUs - in favour of what ?

[Silent_Buddha]

Nice...pretty interesting.

Biggest shrink was Dx10/11 combined with XP. That's a good sign. Less reasons for developers to target DX9 in future projects, especially with next gen consoles presumably 2 years or less away. Unfortunately XP still has 16% of the Steam users still making it an attractive target for added sales. But at least it continues it's very slow decline.

64 bit systems at 60%. Really wish that would get larger. Hopefully, we'll have more developers offering a 64 bit version of their executable. Although without the death of 32 bit I doubt we'll ever really reap the benefits. (Targetted use of greater than 2 GB VA).

Bothersome. 2 CPU systems increased while 4 CPU systems decreased. 1 CPU systems also increased.

You can see the impact of Intel's integrated as both Nvidia and AMD dropped share. Hopefully Intel can greatly increase their performance as they now make up almost 10% of Steam systems making them an attractive lower bound for games.

7970 doing quite a bit better than 7950. Over twice as many 7970's as 7950's. But 7950 now showing the same month to month change at +12%

GTX 680 showing now. But with only a partial month and supply issues it's only at +04% month to month.

Next month's Steam report should have a full months worth of data for it.

I wish they'd show Mac versus PC share.

Regards,
SB

[fellix]

Originally Posted by Davros

* Quad-core CPUs - in favour of what ?

It's probably all those Core i3 flavours (with HT) and the higher-end 2500K (w/o HT) model. I don't think Steam Survey makes a difference between virtual and actual cores, for the stats.

[trinibwoy]

Originally Posted by fellix

It's probably all those Core i3 flavours (with HT) and the higher-end 2500K (w/o HT) model. I don't think Steam Survey makes a difference between virtual and actual cores, for the stats.

But that would inflate the quad-core count not reduce it.

[Silent_Buddha]

And that doesn't explain the rise in single core CPU share. I'm really not sure what to make of it. I haven't followed it much but are AMD's APUs single core? If so that might explain some of it.

Regards,
SB

[pcchen]

Originally Posted by fellix

It's probably all those Core i3 flavours (with HT) and the higher-end 2500K (w/o HT) model. I don't think Steam Survey makes a difference between virtual and actual cores, for the stats.

I think they do, as it clearly states "Physical CPUs." If you open the System Information window in Steam, it distinguishes between physical processors and logical processors.

There are some Bobcat CPU with only one core @ 40nm. However, all Llano have two or more cores IIRC.

[HMBR]

Originally Posted by trinibwoy

But that would inflate the quad-core count not reduce it.

steam detects my Core i3 2100 as
" Vendor: GenuineIntel
Speed: 3092 Mhz
4 logical processors
2 physical processors
HyperThreading: Supported
"

there is also a good number of notebooks/macbooks using dual core sandy bridge with HT fast enough for gaming,

[trinibwoy]

April results are out.

Intel graphics wins big again and English makes a comeback. GTX 680 shows up at +0.22% in its first full month, matching the 7970.

Tahiti seems to be languishing in the last month. Pitcairn and Cape Verde show up at 7800 and 7700 series respectively and aren't doing too hot.

[cal_guy]

That's only for DX11 systems i.e. Vista or Win7 with DX11 installed. Of all the 28nm GPUs only the HD 7970 shows up at 0.40% in DX11 GPU section, so there's seems to be quite a few steam users running XP with their 7970s.

[CarstenS]

Individual results are known to change even after they've been posted, probably Steam running out of steam to update them all at once.

[Silent_Buddha]

Interesting...

Dual core CPU share continues to grow (AMD APUs?).

Interesting that 7970 is the ONLY current generation card to break into the overall video card chart as cal_guy noted.

And in Dx11 systems. It's still 0.36% versus 0.22% for the GTX 680. But adoption the past month has been far higher for the GTX 680. +0.18% versus +0.02%.

7800 series sits at 0.12% of Dx11 systems which matches 7950. And 7700 series sits at 0.08%.

So for current gen Dx11 systems. It's AMD at 0.68% versus Nvidia at 0.22% in Dx11 capable systems. That's less than 1% combined after almost 3.5 months. Not exactly overwhelming for either player and shows Nvidia's inability to release product as quickly as AMD. Which is completely opposite of how it used to be. It also shows just how supply constrained 28 nm appears to be, as demand has been there but supply hasn't generally matched it.

Regards,
SB

[CarstenS]

Originally Posted by Silent_Buddha

Interesting that 7970 is the ONLY current generation card to break into the overall video card chart as cal_guy noted.

Interesting - in the chart I am getting out of steam, it is listed first under the DX11 section (search via CRTL+F) - but as the only current-gen card, yes.

[Kaotik]

There's definately something wrong, and not just a bit, just look at this: (DEC - JAN - FEB - MAR - APR - Change last month)
DirectX 9 Shader Model 2.0 GPUs - 1.24% - 1.09% - 0.98% - 1.98% - 2.21% - +0.23%
DirectX 8 GPUs and below - 1.29% - 1.11% - 0.95% - 2.12% - 2.30% - +0.18%
Both got 1% or over jump in March, and still continue to grow?

[CarstenS]

Yeah, especially since it's contrary to this:
http://forum.beyond3d.com/showpost.p...&postcount=654
where a bias towards older systems in the older results was proposed by the steam quote.

[Alexko]

Originally Posted by Kaotik

There's definately something wrong, and not just a bit, just look at this: (DEC - JAN - FEB - MAR - APR - Change last month)
DirectX 9 Shader Model 2.0 GPUs - 1.24% - 1.09% - 0.98% - 1.98% - 2.21% - +0.23%
DirectX 8 GPUs and below - 1.29% - 1.11% - 0.95% - 2.12% - 2.30% - +0.18%
Both got 1% or over jump in March, and still continue to grow?

I think there's just a large margin of error. Among the top movers this month is Intel G33/G31 Express with +0.07%. I can't imagine that actually being the case.

Steam may be somewhat reliable for the total number of a certain SKU, but not change month/month.

[Silent_Buddha]

Has Steam recently attempted penetration into new markets? If those feature some relatively undeveloped countries that could skew some numbers towards older generations of product.

I have noticed a trend of increasing number of ports of former XBLA titles that would run just fine on many of the older architectures.

Regards,
SB

[sebbbi]

Originally Posted by Silent_Buddha

Dual core CPU share continues to grow (AMD APUs?).

Ultrabooks.

Intel graphics is rising heavily, and all Ultrabooks have dual core (4 thread) 17W Sandy Bridge CPUs with Intel HD3000 graphics. This explains both the Intel graphics and the dual core CPU increases. Intel (and laptop manufacturers) have been promoting Ultrabooks very heavily lately. Intel estimated that 40% of notebooks sold this year would be Ultrabooks... a really optimistic estimate, but seems to be at least partially true based on these Steam survey results.

Bought my wife Asus Zenbook (Ultrabook) some time ago. It has 1600x900 resolution 13.3" display, dual core Sandy Bridge (with HT), a 128 GB SSD drive and weights just over one kilogram. It actually beats my old trusty desktop in CPU benchmarks (both single threaded and multithreaded) and in productivity benchmarks (I have Core 2 Quad Q6600, 160 GB SSD, Radeon 5850) and the Ultrabook CPU consumes only 17 watts... Computers have evolved a lot in just a few years

[Davros]

Originally Posted by sebbbi

Bought my wife Asus Zenbook

wow she has an unusual name

[Silent_Buddha]

Originally Posted by sebbbi

Ultrabooks.

Intel graphics is rising heavily, and all Ultrabooks have dual core (4 thread) 17W Sandy Bridge CPUs with Intel HD3000 graphics. This explains both the Intel graphics and the dual core CPU increases. Intel (and laptop manufacturers) have been promoting Ultrabooks very heavily lately. Intel estimated that 40% of notebooks sold this year would be Ultrabooks... a really optimistic estimate, but seems to be at least partially true based on these Steam survey results.

Bought my wife Asus Zenbook (Ultrabook) some time ago. It has 1600x900 resolution 13.3" display, dual core Sandy Bridge (with HT), a 128 GB SSD drive and weights just over one kilogram. It actually beats my old trusty desktop in CPU benchmarks (both single threaded and multithreaded) and in productivity benchmarks (I have Core 2 Quad Q6600, 160 GB SSD, Radeon 5850) and the Ultrabook CPU consumes only 17 watts... Computers have evolved a lot in just a few years

Ah, yes that would explain alot. And ultrabook users would be just the type of person to get in on many of the more casual gametypes available in Steam as well well heavy graphics processing is relatively unneeded.

Regards,
SB

[HMBR]

Originally Posted by sebbbi

Ultrabooks.

Intel graphics is rising heavily, and all Ultrabooks have dual core (4 thread) 17W Sandy Bridge CPUs with Intel HD3000 graphics. This explains both the Intel graphics and the dual core CPU increases. Intel (and laptop manufacturers) have been promoting Ultrabooks very heavily lately. Intel estimated that 40% of notebooks sold this year would be Ultrabooks... a really optimistic estimate, but seems to be at least partially true based on these Steam survey results.

Bought my wife Asus Zenbook (Ultrabook) some time ago. It has 1600x900 resolution 13.3" display, dual core Sandy Bridge (with HT), a 128 GB SSD drive and weights just over one kilogram. It actually beats my old trusty desktop in CPU benchmarks (both single threaded and multithreaded) and in productivity benchmarks (I have Core 2 Quad Q6600, 160 GB SSD, Radeon 5850) and the Ultrabook CPU consumes only 17 watts... Computers have evolved a lot in just a few years

I really doubt ultra books have such a big impact at the moment, I rather think that is about the growth of notebooks in general for gaming (Intel video card increase supports this), many inexpensive Core i3s/i5s (SB) with the HD3000 which is enough to play source engine games,
also there are many with the likes of GT 540 and other inexpensive discrete graphic cards...


APUs are mostly quad cores at the moment,

[DmitryKo]

I wonder why the PC video card usage details page is not showing any card breakdown statistics as of now... is it supposed to blank out the results at the beginning of each month?

OVERALL DISTRIBUTION OF CARDS		FEB	MAR	APR	MAY	JUN	
DirectX 11 GPUs				35.62%	44.66%	44.50%	45.60%	0.00%	-45.60%
DirectX 10 GPUs				58.58%	47.12%	46.59%	45.78%	0.00%	-45.78%
DirectX 9 Shader Model 2b and 3.0 GPUs	3.87%	4.12%	4.19%	3.95%	0.00%	-3.95%
DirectX 9 Shader Model 2.0 GPUs		0.98%	1.98%	2.31%	2.18%	0.00%	-2.18%
DirectX 8 GPUs and below		0.95%	2.12%	2.41%	2.49%	100.00%	+97.51%
									
ALL VIDEO CARDS				FEB	MAR	APR	MAY	JUN	
Other					25.16%	30.01%	30.12%	30.36%	100.00%	+69.64%

DIRECTX 11 GPUS				FEB	MAR	APR	MAY	JUN	

DIRECTX 10 GPUS				FEB	MAR	APR	MAY	JUN	

DIRECTX 9 SHADER MODEL 2B AND 3.0 GPUS	FEB	MAR	APR	MAY	JUN	

DIRECTX 9 SHADER MODEL 2.0 GPUS		FEB	MAR	APR	MAY	JUN