Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
[EDIT]hmmm.....whats going on at AMD? (and history of 3Dfx Rampage)
http://www.techpowerup.com/reviews/AMD/RX_480/
Quote:Quoting from the AMD reviewer's guide "A lot of work has gone into reducing noise levels for the Radeon RX 480." Sorry, but no, gaming noise levels are bad. The reference card is noisier than every single card released in recent times, and it runs at well above 80°C too. We confirmed the temperature and noise levels with other reviewers, so it's not only our sample.
...
Maximum overclock of our sample is 2250 MHz on the memory (13% overclock) and 1335 MHz on the GPU (5% overclock).

While memory overclocks really well, reaching the Overdrive adjustment limit, GPU overclock is extremely limited and one of the smallest I've seen in my reviews so far. AMD does offer the option of voltage control, but its effect seems extremely limited, with no noteworthy frequency gains in my testing.
...
Using these clock frequencies, we ran a quick test of Battlefield 3 to evaluate the gains from overclocking.
Actual 3D performance gained from overclocking is 5.0%.
...
Idle temperatures are good, because the fan does not stop in idle. During gaming temperatures are quite high, reaching 84°C - which isn't alarming, but doesn't exactly boost confidence in the reference design cooler, especially since it's kinda noisy, too.
...
As mentioned before, power efficiency is improved significantly, mostly thanks to the 14 nanometer FinFET process. These improvements bring the RX 480 roughly on the same performance per Watt level as NVIDIA's last-generation Maxwell architecture. NVIDIA's current Pascal architecture, is still over 70% more power-efficient. This makes me wonder a bit, how NVIDIA managed to gain such huge improvements, while bumping up GPU clock to 2 GHz, and AMD is still running in the 1200 MHz range, with worse power consumption.
...

The weakest point of AMD's reference design is certainly the thermal solution. It doesn't use any heatpipes or other high-tech means to keep the card cool. Rather, there is a big slab of metal, with a copper core that has the blower fan sending air across its fins. As a result we are seeing temperatures of up to 84°C, which has the effect that the card will clock down further to keep cool. On average our card ran 1239 MHz, which is in the upper range of AMD's rated 1120 - 1266 MHz clock window. What's even worse than the heat is the terrible fan noise. While idle noise is fine with 29 dBA (an idle-fan-off feature would still have been nice), in gaming the fan ramps up a lot, emitting 41 dBA noise during gaming (not Furmark). This makes the RX 480 the loudest card launched in recent history, much noisier than for example GTX 1080 (which is almost twice as fast). AMD has mentioned to us that the reference design is deliberately weak to leave room for partners to improve on their custom designs. To me this sounds a bit like "let the partners deal with the problem".

With Polaris AMD is introducing a new overclocking control panel called "WattMan" which has tons of options, including voltage control and several ways to adjust the thermal profile. However, overclocking potential on our sample was very slim. All we managed to do stable was increase the GPU clock from 1266 MHz up to 1335 MHz - a lousy 5% increase, again the worst I've seen for years on a reference board. This is further complicated by the fact that the card will often clock down during OC because it a) exceeds the board power limit or b) runs too hot. If you increase the power limit using WattMan, you'll run into the thermal limit quicker. It does seem that there is a huge spread between GPUs on review samples. I've heard of reviewers who see stock temperatures well below 80°C, while others reach up to 89°C. Assuming that AMD has selected the best cards for press review, retail cards might be even worse, which means higher temps, more noise, lower performance.

The real highlight of the Radeon RX 480 is its pricing though. It starts as low as $199 for the 4 GB version, which also comes with slightly lower memory clocks. In this review we have tested the 8 GB version, which is priced at $239. This ensures both cards claim the price/performance throne of all cards we've tested so far. Especially the 4 GB version seems really affordable, I seriously doubt the extra 4 GB on the 8GB variant will be able to provide an extra 20% performance to make up for the 20% increase in cost. So, AMD has successfully captured the market segment for 1080p cards with the RX 480. This effectively obsoletes all previous AMD cards except for Fiji maybe, which itself has been obsoleted by GTX 1070. Everything faster is owned by NVIDIA's iron-grip on the high-end, where huge margins are possible. Also I expect that NVIDIA will be dropping the price of GTX 970 further to get rid of its inventory, which could entice potential shoppers to go for NVIDIA's alternative, and then there's GTX 1060 coming soon, too.

http://www.techpowerup.com/reviews/AMD/R...re/21.html
Quote:At just $398, about the same price as the cheapest GeForce GTX 1070 you can find, or $478 for a pair of 8 GB cards like we have, the Radeon RX 480 CrossFire is not a viable solution, if you plan to buy two cards upfront. When averaged over all our games it is consistently slower than a single GeForce GTX 1070 at all the resolutions that matter - 1080p, 1440p, and 4K. Instead of buying two cards upfront, you're much better off putting your monies into a single GTX 1070, not just for better performance but also to dodge the spectre of application multi-GPU support, which continues to haunt both SLI and CrossFire.

We averaged games that do take advantage of CrossFire in a separate relative performance data-point than the overall relative-performance. The findings are interesting, when averaged among games that do scale, the RX 480 CrossFire is about 5-10% faster than a GTX 1070. However, only 6 out of 16 tests are taking advantage of the second card.

If you only have money for a single card now, and you want to buy a second card later, then you're in for more than playable frame-rates at 2560x1440 resolution, in games that do scale, and even 4K performance with eye-candy watered down a little. The RX 480 CrossFire loses out its overall relative performance big time, due to the number of games that don't scale well (6 out of 16).

In games that do scale, you're treated with upwards of 85% performance uplift. From our Radeon R9 Nano CrossFire review till now, we see that AMD hasn't really spent a lot of time optimizing CrossFire for more games. We can't fault AMD too much, though, because since then, a lot of games that came out don't support multi-GPU at all, due to engine limitations. There's not much AMD or NVIDIA can do about such games, and this is what scares us about multi-GPU solutions going forward.

We have to give AMD credit where due, for including support for 3-way or 4-way CrossFire, unlike NVIDIA which has dropped support for 3-way and 4-way SLI. The RX 480 supports both 3-way and 4-way configurations, and in any application that can take advantage of CrossFire, not just synthetic benchmarks.

With the advent of DirectX 12, we are promised new multi-GPU rendering modes thanks to Microsoft giving developers more control over per-GPU resource allocation, but we doubt that we'll see widespread use of those techniques. Nowadays, games are developed for consoles first (which are single-GPU), and publishers have little interest in spending a lot of developer time (= money) on adding support for exotic multi-GPU configurations that are used by only a small percentage of their customers.

Some of the latest DirectX 11 games (eg: Just Case 3) and the early DirectX 12 ones, are struggling with multi-GPU support. NVIDIA too, has cut out 3-way and 4-way SLI support (but for slightly different reasons), making us wonder if it's worth the trouble following a multi-GPU upgrade path with mid-thru-performance segment cards such as the RX 480. We rather recommend you spend $398-458 on a single GPU solution.
Reply


Messages In This Thread
RE: hmmm.....whats going on at AMD? - by BoFox - 09-29-2015, 09:48 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-01-2015, 02:11 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 09-29-2015, 11:01 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-01-2015, 02:15 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 09-30-2015, 05:17 PM
RE: hmmm.....whats going on at AMD? - by ocre - 10-02-2015, 08:55 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-06-2015, 02:01 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-04-2015, 08:32 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-04-2015, 08:44 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-07-2015, 01:53 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-08-2015, 06:33 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-09-2015, 10:39 PM
RE: hmmm.....whats going on at AMD? - by ocre - 10-11-2015, 07:44 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-09-2015, 10:44 PM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-10-2015, 05:51 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-10-2015, 12:06 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-10-2015, 12:15 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-10-2015, 06:02 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-12-2015, 10:42 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-13-2015, 06:54 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-13-2015, 11:12 AM
RE: hmmm.....whats going on at AMD? - by ocre - 10-14-2015, 09:05 AM
RE: hmmm.....whats going on at AMD? - by BoFox - 10-17-2015, 12:13 AM
RE: [EDIT]hmmm.....whats going on at AMD? (and history of 3Dfx Rampage) - by SteelCrysis - 06-29-2016, 07:46 PM

Forum Jump:


Users browsing this thread: 1 Guest(s)