1
00:00:06,040 --> 00:00:13,160
by and unlocked Intel fourth gen core i7 or core i5 processor and get a free copy

2
00:00:10,920 --> 00:00:17,880
of Rome 2 Total War click now to learn more the AMD Radeon r9290 when it

3
00:00:16,279 --> 00:00:22,080
launched to turn the graphics card Market pretty much on its ear by

4
00:00:19,800 --> 00:00:29,519
delivering performance competitive with the then $700 plus gtx780 at

5
00:00:26,760 --> 00:00:34,600
$400 there were some issues however power tune is AMD's technology that

6
00:00:31,800 --> 00:00:41,120
balances power consumption clock speed temperature and fan speed so while the

7
00:00:37,239 --> 00:00:44,079
card is advertised to run it up to 947

8
00:00:41,120 --> 00:00:49,360
MHz it and its Big Brother the 290x often runs into thermal limits causing

9
00:00:47,039 --> 00:00:54,000
clock speeds and voltage to scale down so they don't actually run at the speeds

10
00:00:51,399 --> 00:00:57,480
on the Box in games this throttling Behavior could be blamed on a couple of

11
00:00:55,520 --> 00:01:02,960
things one of them is the high heat output of the Hawaii GPU with its mass

12
00:01:00,480 --> 00:01:07,960
6 billion transistors the other is the lackluster reference cooler with its

13
00:01:04,720 --> 00:01:10,200
loud fan that forces end users to choose

14
00:01:07,960 --> 00:01:14,000
between hairdryer like noise or throttled performance now whatever you

15
00:01:12,479 --> 00:01:18,680
want to blame it on there were a few things that were very clear number one

16
00:01:15,680 --> 00:01:21,040
is that the r9290 already performed very

17
00:01:18,680 --> 00:01:25,799
well against the GTX 780 which prompted NVIDIA to drop prices only down to 500

18
00:01:23,240 --> 00:01:30,040
mind you but it was a drop number two is that we won't have a clear idea of what

19
00:01:27,960 --> 00:01:34,560
the card is truly capable of until until we get better coolers in the form of

20
00:01:31,640 --> 00:01:39,920
twin frozer or direct cu2 class coolers from guys like MSI and ASUS even with

21
00:01:37,159 --> 00:01:44,240
custom a cooling though we'd have no way to know for sure whether any thermal

22
00:01:42,119 --> 00:01:48,840
throttling was occurring we've observed both good and bad things about NVIDIA's

23
00:01:46,520 --> 00:01:54,079
GPU Boost 2.0 which is similar to AMD's power tune a good thing about it is that

24
00:01:51,000 --> 00:01:55,640
unlike AMD they provide a base clock so

25
00:01:54,079 --> 00:01:59,240
you'll have a reference point for the minimum performance you'll be getting if

26
00:01:57,159 --> 00:02:05,039
you buy the card one bad thing we've noticed however is that while GPU Boost

27
00:02:01,520 --> 00:02:07,280
2.0 provides enduser customization and

28
00:02:05,039 --> 00:02:13,040
allows you to change your thermal limit manually for example from 80° to 95° it

29
00:02:11,000 --> 00:02:18,480
seems to be actually just doing whatever it wants in the background

30
00:02:15,360 --> 00:02:21,160
so yeah you can change it but it's just

31
00:02:18,480 --> 00:02:26,200
throttling anyway so we wanted to remove all possibility of thermal throttling

32
00:02:23,120 --> 00:02:28,640
and liquid cool the cards fortunately we

33
00:02:26,200 --> 00:02:33,200
contacted klance who hooked us up not only with a water block that fits on

34
00:02:30,480 --> 00:02:36,560
either the R9 290 or the 290x you can check the link in the video description

35
00:02:34,519 --> 00:02:42,560
for pricing and availability of the gear we used in this video but also a GTX 780

36
00:02:40,040 --> 00:02:46,720
block a pump a reservoir and some awesome awesome quick disconnect

37
00:02:44,760 --> 00:02:51,560
fittings that let us do very fast graphics card swaps on our first ever

38
00:02:49,239 --> 00:02:55,640
liquid cooled test bench in fact guys I just disconnected a tube here that's

39
00:02:53,440 --> 00:03:00,080
completely full of water and we got no leakage whatsoever all right moving on

40
00:02:57,480 --> 00:03:06,319
to the test we used the Maximus 5 formul with a 3770k at 3.9 GHz 8 gigs of DDR3

41
00:03:03,360 --> 00:03:12,440
memory at 1600 MHz and a PC power and cooling 1200 W power supply as always

42
00:03:09,599 --> 00:03:17,640
our graphics cards were overclocked as high as they could go on both air and

43
00:03:15,080 --> 00:03:22,200
water and we used two separate liquid cooling Loops for the CPU and for the

44
00:03:20,280 --> 00:03:26,200
graphics card so we could isolate the heat generated by the GPU so we could

45
00:03:23,959 --> 00:03:28,920
remove variables when evaluating them in terms of their thermal performance if

46
00:03:27,599 --> 00:03:32,000
you guys want to see the clock speeds we were running at on the graphics cards

47
00:03:30,599 --> 00:03:37,080
there will be a link to our chart in the video description below our results

48
00:03:34,120 --> 00:03:41,760
surprised us a lot uh we were like everyone on the entire internet

49
00:03:38,799 --> 00:03:46,120
expecting the r9290 to overclock like a beast and destroy the 780 out of which

50
00:03:44,560 --> 00:03:52,079
we were expecting only a mild performance Improvement now because we

51
00:03:48,840 --> 00:03:54,159
traditionally set our 780 to 95° on the

52
00:03:52,079 --> 00:03:59,000
thermal limit for an apples to Apple's comparison against AMD cards and it only

53
00:03:56,360 --> 00:04:02,480
ever runs at 80° we assumed that there was no additional head room to unlock

54
00:04:00,920 --> 00:04:07,560
with better cooling because we were never hitting our userdefined thermal

55
00:04:04,360 --> 00:04:09,760
limits but it turns out that for liquid

56
00:04:07,560 --> 00:04:13,840
cooling enthusiasts NVIDIA's monkey business with the user adjustable

57
00:04:12,079 --> 00:04:18,239
thermal limits turns into some pretty good performance gains for that card

58
00:04:15,959 --> 00:04:23,960
when it's running underwater as well so first up is thermal performance both

59
00:04:20,160 --> 00:04:26,360
cards ran between 35 to 45° C underload

60
00:04:23,960 --> 00:04:30,759
with the 780 being closer to 35 and the 290 being closer to 45 uh when it comes

61
00:04:29,039 --> 00:04:34,440
to gaming performance we only really tested a couple games because what we

62
00:04:32,440 --> 00:04:39,039
were looking for in this video was performance scaling from air to water so

63
00:04:37,160 --> 00:04:44,759
a huge library of games wasn't really necessary what we found is that the max

64
00:04:41,880 --> 00:04:49,360
overclocked performance of the GTX 780 improved by just under 10% and the max

65
00:04:47,600 --> 00:04:54,479
overclock performance of the r9290 improved by about 7% that means that

66
00:04:52,759 --> 00:04:58,320
basically when you're out there shopping and deciding between these two cards

67
00:04:56,400 --> 00:05:03,479
with the intention of going water you can basically look add any performance

68
00:05:00,560 --> 00:05:08,759
comparison chart and factor in another 5 to 10% for each when you're doing your

69
00:05:05,720 --> 00:05:10,759
value analysis on the subject of value

70
00:05:08,759 --> 00:05:14,759
and performance this video is a great example of why I don't talk about

71
00:05:12,880 --> 00:05:20,880
pricing for products in my videos the r9290 launched at $400 which would have

72
00:05:17,800 --> 00:05:24,039
made it a clear winner against the $500

73
00:05:20,880 --> 00:05:26,960
gtx780 in this particular Showdown but

74
00:05:24,039 --> 00:05:31,960
it now costs $500 too making it exactly the same price but with higher power

75
00:05:29,000 --> 00:05:37,800
consumption and higher heat output so for the gamers out there the GTX 780

76
00:05:34,360 --> 00:05:39,360
gets the nod from me at the moment but

77
00:05:37,800 --> 00:05:43,560
there are things other than games and amds cards are well known to perform

78
00:05:41,960 --> 00:05:47,639
incredibly well against their NVIDIA counterparts in some Computer

79
00:05:45,400 --> 00:05:53,039
Applications one example of which being cryptocurrency Mining and folks are

80
00:05:50,560 --> 00:05:56,680
buying up AMD graphics cards by the truckload for Bitcoin and Litecoin

81
00:05:55,080 --> 00:06:01,080
Mining and that's been causing some serious changes to the retail pricing as

82
00:05:59,039 --> 00:06:05,880
I outlined before so ultimately it'll be up to you the viewer to look at the

83
00:06:03,160 --> 00:06:09,000
performance numbers then check out the values that we're showing you in terms

84
00:06:07,120 --> 00:06:12,120
of what you can expect when overclocking then check out the links to the cards in

85
00:06:10,479 --> 00:06:15,880
the video description and do the value calculation for yourself if you're

86
00:06:14,160 --> 00:06:18,400
curious guys you can see the overclocked values for our graphics cards in the

87
00:06:17,240 --> 00:06:22,680
Google Doc Linked In the video description remember that overclocking

88
00:06:20,560 --> 00:06:26,240
is always a crapshoot and you may get a card that overclocks better than ours

89
00:06:24,639 --> 00:06:30,080
you may get one that overclocks worse than ours and there's no guarantee for

90
00:06:28,520 --> 00:06:35,440
any of that stuff so anyway guys I hope you enjoyed this video on the removal of

91
00:06:32,639 --> 00:06:40,360
thermal limits from the r9290 and the gtx780 let us know in the comments if

92
00:06:37,919 --> 00:06:43,840
the results surprised you or if they were what you expected like the video if

93
00:06:42,440 --> 00:06:47,639
you liked it dislike it if you disliked it and not as always wow it's doing

94
00:06:46,039 --> 00:06:52,180
really good up till there as always don't forget to subscribe to lonus Tech

95
00:06:49,599 --> 00:07:15,470
tips