Q: max-OC'd FTW GTX460 vs max-OC'd XFX Vapor Chamber 6850, which wins? Answer inside.

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
Your way off about power costs, and I am very familiar with the electric utilities also.
How many hours a day to figure people game/full load. 1 hour a day ? 365 days a year ? http://michaelbluejay.com/electricity/howmuch.html
Its practically nothing. I just changed my bathroom vanities and went from 3-60 watt bulbs to 4-60 watt bulbs.
I know that will effect my bill more, than 25 watts at full gaming load.
??
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
Your way off about power costs, and I am very familiar with the electric utilities also.
How many hours a day to figure people game/full load. 1 hour a day ? 365 days a year ? http://michaelbluejay.com/electricity/howmuch.html
Its practically nothing. I just changed my bathroom vanities and went from 3-60 watt bulbs to 4-60 watt bulbs.
I know that will effect my bill more, than 25 watts at full gaming load.
??

Show me how my math is wrong? Note I was calculating idle power draw differentials, not load, which is what you seem to be thinking. Re-read what I wrote. It was Happy who talked about load only; I talked about idle AND load.

As for 24/7 usage, I know lots of people who leave their PCs on 24/7. Personally I don't (surprise, surprise), which is why I didn't think TOO much about it when I was weighing each card's price/perf. The GTX460-1GB vs. HD6850 price difference is more like $5-10 for someone with my usage habits.

But different people have different habits, different # of hours they game, different efficiencies in PSUs, and different electricity costs, which is why I didn't want to get too much into that in the first post of this thread. I mentioned power but didn't make a big deal out of it until Happy went off on me a few posts ago. He was wrong about how much power can cost, so I corrected him. That's all. And you can correct me if *I* am wrong, too. Show me the math, though.

Your light bulb thing is a false argument because you can save the same amount of power whether you use a GTX460 or HD6850. Or can you *only* switch to more-efficient appliances if you own a GTX460?

P.S. If your argument is more that you should concentrate on higher-potential-savings appliances first, I agree with you. But $17 is significant compared to the cost of a midrange video card and thus can alter price/performance ratios, which is what we were talking about in this thread. If we want to change the thread topic to something else, we can (perhaps by creating another thread in another forum), but I'm just talking about video card price/performance factoring in power costs.

And if you want retail residential power prices rather than a map, try: http://www.eia.doe.gov/electricity/epm/table5_6_a.html

Note the avg in't weighted by population. Looking at the big-pop states (http://en.wikipedia.org/wiki/List_of_U.S._states_and_territories_by_population) it could be more than 12.01 cents/kWh; if you care enough you can run the #s for us for ALL states, weighted by pop.

top 10 states by pop:

population (MM=million) state cents/kwh

37MM ca 15.52
25MM tx 11.99
19.5MM ny 19.58
18.5MM fl 11.68
13MM il 12.00
12.5MM pa 13.34
11.5MM oh 11.72
10MM mi 12.86
10MM ga 10.85
9.5MM nc 10.41

Eyeballing it though you can see the little states pay less on avg: http://www.green-energy-efficient-homes.com/how-much-does-electricity-cost.html

Note also that we're in a huge recession and thus input costs are lower than they were in 2008 prior to the recession start, thus these numbers may look low once the recession lifts.
 
Last edited:

shangshang

Senior member
May 17, 2008
830
0
0
^_^ LOL $17 is not significant if you're a gamer. it costs me $40/week for gas just to commute to work in a Toyota sedan. I'm sure others in SUVs easily would spend $80-$100. $17 is money, and money is money, but i'm somewhat amused at the attempt to inflate $17.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
^_^ LOL $17 is not significant if you're a gamer. it costs me $40/week for gas just to commute to work in a Toyota sedan. I'm sure others in SUVs easily would spend $80-$100. $17 is money, and money is money, but i'm somewhat amused at the attempt to inflate $17.

You are changing the topic by saying it's no big deal in other contexts. The topic of this thread is comparing the GTX460 vs 6850, and one thing that keeps popping up is price/perf comparisons. $17 can be significant when comparing $180 products' price/perf when they are close in perf. Note also that the $17 is a recurring cost and is >$50 after 3 years, for instance. Just something to think about. I still think it's hard to make a bad choice between a GTX460-1GB and a 6850.
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
TechReport managed a 940mhz/1175mhz memory clock on the MSI GTX 460 Hawk Talon Attack. Temperatures still remained at 72*C. The XFX 6850 managed 950mhz GPU but at those frequencies, the card operated at 84°C, with the fan at 90% of peak. Once they pushed the 6850 to 1.3V and higher GPU clocks, the temps were near 90°C.

Looks like GTX460 is able to overclock, while retaining reasonable noise levels and temperatures, unlike the 6850. It would be more fair to wait for aftermarket coolers on 6850s to compare apples to apples imo. Stock coolers on 6850s are too inferior for a fair comparison. And the MSI card costs $214 on Newegg which is quite a bit more $$ compared to $179 HD6850s.

The power consumption on the 460 approaches GTX480 at 940mhz overclock though. :rolleyes:

avp1920.gif


powerload.gif
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
TechReport managed a 940mhz/1175mhz memory clock on the MSI GTX 460 Hawk Talon Attack. Temperatures still remained at 72*C. The XFX 6850 managed 950mhz GPU but at those frequencies, the card operated at 84°C, with the fan at 90% of peak. Once they pushed the 6850 to 1.3V and higher GPU clocks, the temps were near 90°C.

Looks like GTX460 is able to overclock, while retaining reasonable noise levels and temperatures, unlike the 6850. It would be more fair to wait for aftermarket coolers on 6850s to compare apples to apples imo. Stock coolers on 6850s are too inferior for a fair comparison. And the MSI card costs $214 on Newegg which is quite a bit more $$ compared to $179 HD6850s.

The power consumption on the 460 approaches GTX480 at 940mhz overclock though. :rolleyes:

avp1920.gif


powerload.gif

HAWK has great cooling. I came so close to getting the GTX460-768MB w/ Twin Frozr II cooler at an effective cost of $137 (after cc and other discounts) for 1680x1050, after it showed up just a few days ago (and sold out in a day), but ultimately I wanted Eyefinity and will do my big spending at 28nm rather than 40nm, so I settled for a 6850 as a stopgap measure till next fall. But the GTX460-768 w/ Twin Frozr II would have been the ultimate bang for the buck card. They don't make Twin Frozr IIs anymore, they only make HAWK cards now (at a higher price), but I can't see any difference in the hardware other than allowing you to individually tweak component voltages.
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
The 6850 is an awesome card, esp. for a stopgap coming from the 4350 or something you were using last week?
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
The 6850 is an awesome card, esp. for a stopgap coming from the 4350 or something you were using last week?

No, the whole sordid tale I've repeated earlier this thread. Basically I thought I could give up Eyefinity, and as a matter of principle I didn't want to pay price-gouging prices, so I got a EVGA GTX460-768. But the card shipped with a resistor that fell off. I RMA'd it (but am still going to use it for the next 2 weeks since Amazon is giving me 30 days to return it) and ordered an MSI replacement but 2 days later a) it still hadn't shipped yet (not even STARTED to ship!), b) I realized how much I missed Eyefinity after all, and c) 6850 prices fell to normal. So I canceled the MSI GTX460 order and ordered an ASUS 6850 though it says it won't ship for another 2-4 weeks. In the meantime I saw the GTX460 Twin Frozr II 768MB for sale and was tempted... if I had only one or two monitors I would DEFINITELY have gone for it. I also read in-depth reviews of the ASUS that makes me think they are cheaping out with it. They added a power phase with the GTX460 but none for the 6850, for example. So I may end up canceling the ASUS in favor of a Sapphire 6850 which is supposed to have excellent cooling too, just no PCB stiffener, but I'm not sure if that's a big deal or not. So now it's down to ASUS v. Sapphire 6850 for me. It really depends on what Anandtech's 6850 roundup says about the ASUS vs. Sapphire. In any case the EVGA 460-768 buys me a few weeks' time to decide... I am no longer on IGP, just on a defective card that may crap out at any moment. :)

P.S. Under no circumstances should ANYONE ever buy a HIS 6850; they have a pathetic heatsink/fan: http://www.techpowerup.com/reviews/HIS/Radeon_HD_6850/4.html
 
Last edited:

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
I believe he was using the IGP :p

Believe it or not, a HD4250@1GHz can give you semi-playable framerates in Left4Dead at 1280x800. Newer IGPs might be able to do better. And Source is a pretty efficient graphics engine, apparently.

This probably also explains why even a 6850 can run L4D at 5040x1050 at 60fps. I wouldn't want to try Eyefinity with anything newer than Source-based games. There is a rumor that Half Life 2 Episode 3 will use a whole new engine, since it's been delayed so long. :)
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Going to pick up a 6850 when I see some better aftermarket coolers.

Definitely! More and more reports are coming out that HD6870 is a poor overclocker. Most can't go beyond 1000mhz. So it makes perfect sense to save $60 and grab an HD6850 instead if you plan on overclocking. A 900-950mhz 6850 will be within 10% of an HD6850 @ 1000mhz, but $60 is a lot of savings.

Check out Gigabyte 6850 for $189.99.

P.S. Under no circumstances should ANYONE ever buy a HIS 6850; they have a pathetic heatsink/fan: http://www.techpowerup.com/reviews/HIS/Radeon_HD_6850/4.html

That massive shroud cover on the HIS card is like a padded bra.
 
Last edited:

notty22

Diamond Member
Jan 1, 2010
3,375
0
0

Ares1214

Senior member
Sep 12, 2010
268
0
0
I disagree with the Asus being a bad cooler, actually very good, however if you really think its bad, you would have to pay more. IE, get the MSI R6850.
 

Xzyrus

Junior Member
Sep 25, 2010
22
0
0

See my post here regarding 6850 coolers:
http://forums.anandtech.com/showpost.php?p=30670612&postcount=6

Gigabyte's dual fan might explain better performance than the ASUS DirectCu. The Sapphire cooler which keeps up with the Gigabyte one doesn't look half bad either:
http://www.pro-clockers.com/videocards/1696-sapphire-radeon-hd-6850.html?start=2

From my other post, they both seem to edge out the ASUS DirectCu.

Regarding the topic, raw power = GTX460, but at 169.95 CAD, I couldn't resist the 6850!
 

MangoX

Senior member
Feb 13, 2001
594
108
106
Definitely! More and more reports are coming out that HD6870 is a poor overclocker. Most can't go beyond 1000mhz. So it makes perfect sense to save $60 and grab an HD6850 instead if you plan on overclocking. A 900-950mhz 6850 will be within 10% of an HD6850 @ 1000mhz, but $60 is a lot of savings.

Check out Gigabyte 6850 for $189.99.



That massive shroud cover on the HIS card is like a padded bra.

Oh my that cooler looks terrible! Like the same size as a stock Intel i5 HSF.

Luckily I'm in no rush. My new build is coming together slowly so I don't need a card for a while.
 

Arkadrel

Diamond Member
Oct 19, 2010
3,681
2
0
powerload.gif


Only 123 watt more used by the 460 card than the 6850.
Still damn impressive that a 460 can reach that level of performance with a overclock.

so a 460@ 940 uses 123 watts to get 13% more performance than a 6850 thats overclocked. How much does 123 watts come to if you play 5 hours a day, on a yearly base?
 

happy medium

Lifer
Jun 8, 2003
14,387
480
126
so a 460@ 940 uses 123 watts to get 13% more performance than a 6850 thats overclocked

Your math is off. 388 watts vs 324 watts (65 watts more) and at that overclock its more then a 13% performance difference.

Take the gtx460 ftw @ 850 core's wattage (317 watts) vs the 6850's 950 core's wattage (324 watts). A gtx 460 @ 850 core beats a 6850 @ 950 core. Just look at the original review in thie op's post.
 
Last edited:

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
so a 460@ 940 uses 123 watts to get 13% more performance than a 6850 thats overclocked. How much does 123 watts come to if you play 5 hours a day, on a yearly base?

GTX460 idle watts are 15w higher and probably matter more than load watts since most people don't game more than a few hours a day but leave their PCs running 24/7. That said, if the delta is more than 15w at load that means the $17.08/year delta in cost is an underestimate. That's why I said >17.08.
 

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
All depends where you get your data from, and what card you compare it to.
Nvidia has very polished bios's and they usually have the best idle power numbers/behavior. That includes in their o/c states.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
All depends where you get your data from, and what card you compare it to.
Nvidia has very polished bios's and they usually have the best idle power numbers/behavior. That includes in their o/c states.

Then you agree that my math is accurate, assuming HWC's idle power estimates were correct.

I already said that power draws might vary, earlier.

You see why I didn't want to get into this whole power discussion in the first post of this thread? That's why I mentioned it but didn't go in-depth into it in the first post of this thread. AIB power draws might vary, or PSU efficiency may vary, or a number of things. But from what HWC's review indicates, it's a 15w delta at idle. (Actually I checked again, it's only 14w delta compared to the reference card, my bad. But that's still ~$16/year, year after year, instead of ~$17. And if load deltas are >14w then we're back up to $17 or more, per year.)

Since the total cost of ownership is higher for the GTX460, but it's also potentially higher-performing as well, I think the easiest thing to do is to figure that GTX460 = HD6850 in price/perf. Then buy based on features (e.g., CUDA, Eyefinity, PhysX), or whatever floats your boat (color or design of cooler, etc.). (But if you live somewhere where electricity is much more expensive than in the U.S., then perhaps the 6850 is a wiser choice.)

In other words, I stand by the first post of this thread: it's hard to go wrong either way. Just buy a GTX460 or HD6850 and enjoy! :)
 

notty22

Diamond Member
Jan 1, 2010
3,375
0
0
No , I don't agree with your other post.
I'm not going to argue the point. I will explain why if its not obvious later.
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
No , I don't agree with your other post.
I'm not going to argue the point. I will explain why if its not obvious later.

I look forward to your attempt to disprove simple math if we assume HWC was measuring accurately.

If instead you want to argue that the idle power draws are equal, feel free. I have seen some reviews that claim that. So perhaps the number is between 0 and $X dollars ($17 is for 13 cents/kWh @ 24/7). *shrug* It varies depending on AIB and measurements, yes. But given my assumption that HWC measured accurately, then I will go with that.

Note: it could be the case that the GTX460 card draws equal wattage to the 6850 hence why some reviewers found that they tied, however, maybe the cooler pushed more hot air back into the case or something which necessitated system fans to spin faster, hence the higher *systemwide* power draw.

Another possibility is drivers. Tom's did a study recently showing how CPU utilization could be higher or lower depending on whether an AMD or NV card was running. So perhaps NV drivers eat up more CPU resources so that the overall system power draw is higher for NV cards?

P.S. See here: http://www.guru3d.com/article/radeon-hd-6850-6870-review/10

13 watt discrepancy, and not only that, they also talked about euros/kWh price. Probably because Europeans pay more for electricity on average, so it's a bigger deal there. Stock volts, though, unlike HWC's higher volts.

Other reviews show that the difference is more like 4-6 watts at stock volts. http://www.hardocp.com/article/2010/10/21/amd_radeon_hd_6870_6850_video_card_review/8 http://www.bit-tech.net/hardware/graphics/2010/10/22/ati-radeon-hd-6850-review/9

There could be other explanations for the 15w discrepancy. I'm not saying this is an exhaustive list of possibilities. Whatever the may be, it's clear that the 6850 has a slight power advantage and performance disadvantage when both the GTX and Radeon are clocked high and overvolted.
 
Last edited:

happy medium

Lifer
Jun 8, 2003
14,387
480
126
I'm gonna make a new thread for this subject with a poll.

Titled
Would you rather have 15 less watts or 15% more performance in a video card for the same price.

Hows that Blastingcap?
 

blastingcap

Diamond Member
Sep 16, 2010
6,654
5
76
I'm gonna make a new thread for this subject with a poll.

Titled
Would you rather have 15 less watts or 15% more performance in a video card for the same price.

Hows that Blastingcap?

I would phrase it differently if your point is price/performance. You are clouding the issue by phrasing it that way. Why not make this poll instead:

Which Three Year Ownership Plan would you prefer:

1. HD6850 at $180 without rebate.

2. GTX460-1GB at $180 after rebate, plus an additional $50 in electricity costs above and beyond the HD6850.

3. Neither because this poll was set up to air some sort of weird beef Happy Medium has with Blastingcap despite Blastingcap not knowing or caring who Happy Medium is, other than his continual personal attacks as to Blastingcap's character.

You may also find disagreement that the GTX460 is 15% faster in actual gameplay situations rather than canned benchmarks (see, e.g., the HardOCP review).

One thing I wonder: why is this so personal with you? I don't know you, you're just some dude in Philly. I would prefer it if you would be less personal, like swearing at me and hurling insults like "sunk to a new low" which implies that I was low to begin with and am low now.. for what, point out power draws? My math was accurate. Read my OP in this thread. I think it was a fair summary. RS gave me kudos for the suggestion at the end of it, too.
 
Last edited: