KitGuru KitGuru.net – Tech News | Hardware News | Hardware Reviews | IOS | Mobile | Gaming | Graphics Cards
This week we saw the official announcement of Nvidia's latest addition to the Pascal family, the GTX 1060. At the time of announcement we got the specifications and we learned a little more about Pascal's exclusive features but we did not get pricing information for those of us in the UK. Fortunately this information has now been tracked down and we have confirmation on Nvidia's UK RRP.
Nvidia has promised that we will see custom cooled cards on the 19th of July but once again, there will also be a run of Founder's Edition cards, which will have a higher RRP, much like the GTX 1080 and GTX 1070. However, for the GTX 1060, Founder's Edition cards will be limited edition.
Here in the UK, the GTX 1060 Founder's Edition will cost £275 and will be available through Nvidia's own store exclusively for a limited time. Custom cards from partners like ASUS, MSI, EVGA, Zotac, Gigabyte and more will start at £239.99 though as the GTX 1080 launch showed us, there may not be many custom boards sticking to the RRP at all, or launching below Founder's Edition prices.
Demand could also be an issue, GTX 1080s and GTX 1070s have been going out of stock frequently since launch and pricing has fluctuated quite a bit as a result so we will have to wait and see how things play out following the 19th of July, when the cards hit retail.
KitGuru Says: Nvidia's Pascal graphics cards have been in high demand since launch so while Nvidia's recommended retail pricing gives us a rough idea of what to expect on launch day, there is no guarantee that things won't change. Are any of you looking to buy a new GPU this summer? Are you considering the GTX 1060?
Not me, but I shall be helping a friend upgrade her GPU in a few weeks time.. and the 1060 is probably going to be the one. Waiting for proper performance tests first though.
You should definitely wait for the other AMDs and see how they fare, they will likely be a nice competitors since rx480 is already a very aggressive with its pricing.
I heard some people said AMD’s GPU raw power is greater than NVIDIA’s. Even if that’s true, AMD needs to work seriously on unstableness issues.
249 USD = 239 GBP? Pretty sure exchange rates haven’t gone THAT bad…
Then again I guess it’ll be like the 1080, supposedly 599 USD, but then there are some on sale for 650 GBP… sure is shit to be living in the UK for anything tech related.
540 USD for a i7-6850K? Nah, try 550 GBP.
If they sell the 1060 for $299, they just forfeit their 480 counter. Nobody is going to spend $299 on something they can spend $199 on.
Well currently, the 1060 is nore efficient, a little faster, and will probably retail around a similar price (or a llittle higher) to the RX480 in the UK.
The next things were seeing from AMD are the RX470 and RX460. Neither of which will provide any kind of benfit over the 1060 except being cheaper, and even slower.
If rumours of an RX490 being a dual-gpu card are true, then that may aswell be disregarded also, as the price will be irrelevant do this kind of price/performance area.
Totally incorrect ; )
RX480 will be easlily cheaper and equal in DX11 while much better in DX12. And custom oced ones will be clearly faster and a bit cheaper than 1060 too. Only thing against RX480 is power consumption by 30W.
well, leaked benchmarks say otherwise. Still, you can buy a 480 today, while a 1060 will be… soon, and then more expensive than suggested, and possibly out of stock too.
Well, except fanboys…
You forget to remember the fact that the 1060 can be overclocked too.
Benches have already shown the 1060 to be ~15-20% faster.
And once again the DX11 vs DX12 thing.. you’ll find that Pascal does have async and does not suffer in DX12 the way that Maxwell does.
What i find impressive is that 1280 cores in the 1060 is outpacing the 2304 in the RX480.
Pascal was built around the new process node and pushing the efficiency and clockspeeds way beyond previous architectural limits. Wheras Polaris, well that’s just an advancement of the GCN architecture on the new process node. Slightly dissapointing to me.
AMD has control of the entry level area of the market currently, I hope they can come back and compete with their bigger chips, Vega or whatever it will be. And maybe by that time we will see the next GTX Titan and 1080Ti. Thats when i’ll be getting interested in upgrading my Titan X.
its yet another reason to get out of this shit country
8gb vs 6gb, better asynchronous compute, better DX12 implementation vs better efficiency and more raw power. Typical Nvidia for better immediate benchmarks where the AMD card will be out performing it two years down the line. I’ve had the same card for the last three and a half years and its still performing strongly.
The difference between the RX 480 and the GTX 1060 seems to be quite dramatic. At this moment a RX 480 gets you a GTX 970, whilst the GTX 1060 is claiming to have the performance of a 980.
But to be fair, AMD did claim to have a 980’s performance with the RX 480, so we won’t know until performance benchmarks come out.
Shit all over Europe.
Well it’s obviously not just exchange rates it is important taxes and shipping costs. Something UK and Europeans customers seem to be forgetting.
Don’t let the door hit you on the way out
No you are totally incorrect, the cost of a RX 480 in Europe is a min £240+ and prices average around £250. It uses more than the 150watts it was advertised and has the performance of a gtx 970 where as the 1060 will have the performance of a 980 and use less power than both a 980 and an rx480 and cost even less assuming prices stay as advertised. You would be an idiot to choose an Rx 480 right now.
It’s always been like that, for at least 20 years anyway – if you want UK pricing you swap the $ for a £. The thing is, the US prices don’t include taxes, so while it seems unfair, it’s less a kick in the bollocks and more a headbutt to the nose.
Pre-2001 if you had US friends in states with low sales taxes you could get em to buy it for you and ship it. Nowadays though you’ll be super lucky not for it to get caught at Customs and have to pay duty on it
<<o. ✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:::::::!uf115w:….,…..
Uhm… no. The 480 equals an overclocked 970 at stock which is basically in line 980 performance.
The 480 can also be additionally undervolted at the core and VRAM by quite a bit (because AMD uses higher voltages on their GPU’s), allowing the GPU to sustain it’s boost clocks for long periods of time and it also reduces power consumption to about 120W or less.
So, performance/efficiency-wise, the 480 and 1060 will basically be in line with each other as far as DX11 performance goes (unless 1060 turns out much faster or much slower in real games) with 480 having a distinct edge in DX12.
But 480 is ALREADY cheaper and has pretty good software support from AMD which is fast and of higher quality than before.
480 can also be undervolted (due to higher than usual voltages that AMD applies), so power consumption can be slashed by 30W, which also raises performance of 480 too by about 4-5% (because the boost clocks can be sustained without issues at that point).
Nvidia already optimized their GPU’s for lower voltages, hence why they appear more energy efficient.
Personally, I prefer to undervolt manually as it gives me some breathing room and I can probably both undervolt and overclock at the same time.
No benchmarks have actually been released and as far as the charts Nvidia leaked much like AMD’s leaked charts they are always cherry picked!
To clear something up… 480 has a performance level of overclocked 970, which is basically in line with 980.
Also, AMD cards come with higher voltages by default… which can be easily lowered using Wattman.
Reducing the voltage on 480 core reduces power consumption by over 30W, and there’s also VRAM that can also be undervolted by as much as half in some cases (so the overall result would be about 45W less power draw at stock levels).
This undervolt results in performance increase of about 4-5% because the boost clocks can be maintained for long periods of time, and the recent drivers provided an additional 5% boost in performance.
Nvidia already tunes their GPU’s to lower voltages, that’s why they appear more efficient and faster.
AMD uses higher voltages most likely to improve yields.
Realistically, it depends on how 1060 performs in real games, but I don’t think the differences will be as drastic as you claim.
Even more-so, at DX12, I don’t think Nvidia will necessarily have an advantage of any kind, especially at 2K (whether DX11 or DX12).
Plus, 480 is cheaper.
before speaking we should wait and watch the price of 1060 in europe, you guys didn’t learn anything from last month… it will be overpriced for a lot of time. Also we don’t know the exact price of customs, and the “reference” is the fe edition who arrives at €300+.
the 1060 dx12/vulkan performance will be a key factor. the 480 comes close in some dx12 titles to the 1070. If the 480 beats the 1060 in dx12 i might just get a 480
The only reason I’m buying the 1060 over the rx 480 is because of nvidia shadowplay.
AMD also has something akin to Shadowplay:
https://actionrecorder.com/en/tutorials/hardware-settings/amd-app-hardware-acceleration-for-game-recording-and-live-streaming
The RX 480 costs as much as $500 at Amazon.com. I’ve yet to see a single $200 card, and though I saw one $240 in stock, that was for a very short time (pointed it out to a friend, who tried to order it some hours later and it was no longer available).
If NVIDIA can has a good stock of 1060’s before RX 480 prices drop to normal, it will be great value at $300 by comparison.
Here but OC’ed can do even better 12k easy.
Here we have OCed XFX with Backplate 1330MHz for ~269€
Bench on reference non OC RX 480 from Guru3D
http://i67.tinypic.com/2u3upm0.jpg
GTX1070 still being price gouged a month after launch so don’t expect much difference from the 1060, especially if it is a little faster than an RX 480.
I’m in the market for a new card right now but I’m not paying extra for Nvidia’s (or TSMC’s) inability to produce enough for demand. I’ll just wait, and let the market calm down because it’s still red hot right now.
It might be another month or even two, but I’m not in a massive rush. We’re not exactly in a flood of new games in the summer, they won’t start until September time typically.
Yeah, Plays.tv is a piece of fucking trash. They need to get some devs on it now if they expect to compete with Nvidia.
Radeons have always had more raw oomph then competing Geforce(as well as much better build quality i.e VRM proper spec etc) this is why they make tremendous crunching machines, however, they also take more finesse to get the potential out of them, main reason they tend to use a bit more power then we would expect, they have much more “in the core” that needs power, while Nvidia is/has chopped things out cause “dont need it” which has helped reduce their power though in many cases, not nearly as drastic as many review sites point out, Radeon more like a diesel whereas Geforce more like a show car, one is meant for grunt while still being able to race efficiently other meant for speed, limited trunk size and hampered by knock off quality parts ^.^
Also all those saying 10 series more efficient, maybe at moment sure, but this is also very new ground for Radeon using that state of the art voltage design, this can be helped via downvolting and I am quite sure give a month or few performance will be quite a bit higher and power noticeably lower once they had time to optimize it, whereas GTX1060 is likely to NOT do this, otherwise it castrates sales of higher ASP cards, Nvidia likes $$$$$.
Pascal does NOT have Async this is AMD specific like PhysX is for Nvidia. Pascal effectively just chopped things away and added a few others compared to Maxwell and heavily pushed its limits which likely means if they used underspec components as they usually do, they will not last long, take them and declock some, almost certain Maxwell will actually be slightly faster at similar clocks and touch more power.
You need be choosy of review sites, many of them paid for by Nvidia i.e biased, AMD would be efficient as well(if not more so)if they chopped everything out and relied on gimmicks. DX12/Vulkan AMD shows advantage, wonder why that is, yay GTX 10 series is faster in DX10/11, so basically they chop things out and heavily tune towards older API and charge a premium for gimmick features instead of building for modern API, oh wait that comes next year, so they can reduce performance till then of current cards, make the new ones look “modern” then charge a premium on them as well lol @ nvidia slave boys.
likely via the way Nvidia builds theirs, there is very little breathing room for “control” they might use lower volts, but how much amps are they using, likely they are using underspec VRM jacking the amps up and keeping volts lower, they are efficent as they tune them(so likely many cannot be sold as they fail cause of very tight acceptable tolerance otherwise there would be no “shortage) as well they chop a bunch out to artificially make them appear faster then older generation which they constantly nerf performance down to upsell new one and say “its X faster”
Nvidia shenanigans, and fanboys drinking that koolaid ^.^
That’s what I meant by Nvidia optimizing their own GPU’s for lower voltages.
They operate on lower voltages by default giving them very little if any breathing room for manual setup – I guess having more money at their side helps them doing that – plus with a larger market share, they don’t feel threatened.
AMD just sets a higher voltage by default so they can churn out more reference models (at least in the case of 480), and this is why the 480 ended up drawing more power than usual in some cases (note, even official reviews noted that some reference 480 gpu’s were affected, not all – and indeed, some people have reported power draw in the area of 110W to 120W).
In cases where power draw was larger than expected, undervolting the 480 to its lowest levels on nothing but the core reduced power consumption by over 30W (or more) and brought it in line with what was initially spoken of the GPU.
This resulted in the 480 being able to sustain its boost clocks effectively for as long as you stressed the gpu and increasing overall performance.
480 is already as efficient as 1060 on both power draw and performance (even more so at DX12) – and in cases where reference 480 draw more power, they can easily enough be undervolted.
The VRAM on 480 can also be undervolted by quite a bit – and someone mentioned they did it – but I do not recall how much in terms of power savings it resulted (likely in the area of about 5 to 10W).
So to those talking about 480 over 1060.. is the 480 power issue solved then ? Or is it still putting your motherboard on a risk of frying itself?
Nvidia can pay off review sites and have professional “shills” to make their stuff always appear better, even though many many many times over the years their quality has been anything but great, but when your stock trades at 8x or more and market cap is so much higher, they can afford to do so, and of course many who buy them dont care, are uninformed etc. its ok, this was last year Nvidia had “deals” with Intel. Not sure what AMD plan is/was, but IMHO they should have called them RX450-460-470 not 460-470-480 this way here they had room to maneuver a bit more.
Nvidia knows the voltage controller they use cause they have been using them fancy ones for a few generations now, whereas for AMD it was I believe Kaveri and 285/380 they were using a preliminary one to current used on the RX 480 which is top notch controller, but also going take them a bit to really know it. That probably all it is, they wanted to make sure as many could be out door as possible with as little tuning as possible to keep costs down a bit, least it shows Polaris is quite tuneable, and likely in a few months time will be quite substantially different then what we see today, it is known from Apple that TSMC 16nm is slightly better then Samsung 14nm though now they have more numbers out, they will be able to learn the ins and outs of the design, and likely will see more out of it seeing as they have AMD/IBM/Samsung/GF all working on it, vs just TSMC, so, what may not look great right now, could turn into night and day difference.
For me, what a could 2x raw power and ~20w less power then my 7870 which I love is, I cannot say anything but love for the RX 480, just waiting on $ to be more stable before I get one, wish they focused on making reference blower better though as am quite partial to the card taking care of its own heat and not blowing it into the case. they have quite a few custom RX 480 that are sexy as hell 🙂
<<o. ✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:✸✸✸✸✸:::::::!uf165w:….,…..
Have NV not reduced the GPGPU performance to obtain the high clocks in Pascal? How do the NV 10 series cards compare to AMD on “typical home compute” uses e.g. photo editing and such? I’d like to see more benchmarks on those kinds of things in reviews.
Yes, fixed in the latest driver.
Yeah I’ve used it before, it kept bugging out on me and randomly stopped recording when I didn’t want it to.
GK104 / GM204 / GP104 have all followed the same route with very little compute power built into them. 1/16th or so.
GPGPU is only hurt significantly if you’re running heavy FP64 work, in which case you’d know to either buy an AMD card or go for the prosumer type cards from either vendor.
As for FP32… both sides have plenty of that.
Async is not a proprietary feature, its an optional spec of DX12. Maxwell was not geared towards DX12 and had no Async anyway. Let’s not forget that AMD cards arent missing out some parts of the DX12 spec too.
Pascal doesnt chop anything out compared to Maxwell, im guessing you read an article about how Pascal ‘eliminates’ things that prevented clock speeds from scaling well. While technically that statement is true, its false, it was more of a rejuggle of the rendering pipeline to get more throughput out of each core. Instead of having to wait for a texture, the optimised caching/compression algorithms will have already gotten that texture ready, not only allowing it to get through that work quicker, but it keeps the load more steady on each core, something thats fairly important when it comes to high clocks and power efficiency.
Im glad that AMD is fixing the issues with the RX480, though i feel they could have done themselves a great deal more good if theyd simply gone the way nVidia does and kept the voltages lower in the first place. I mean, something almost twice as fast used less power? Mhm, doesnt look too good. Not to mention the fact the rest of the system would be pulling more power in order to keep up with that faster card aswell.
We will most likely see nVidia have much better async in the next architecture, whatever that will be.
I want to see a real advancement in AMDs gpus and architectures.
7xxx
8xxx (mainly oem)
Rx 2xx
Rx 3xx
R9 Fury / X / Nano
RX 4xx
Are all pretty much the same architecture with a few tweaks.
We can see how its the same architecture with how the clockspeeds have hit the same wall of around 1250 ish (+/- a bit) until now the process has been dropped to a lower one resulting in slightly improved clocks.
I want to see something new from AMD. Zen, Volta, just something new… not an older architecture with some tweaks n stuff to make it look new. I mean, in the time AMD has tweaked 1 architecture, nVidia has paced its way through 3 architectures which have each made quite significant advancements.
If I seem like an nVidia fanboy to you, then by your own judgement standards you must be an AMD fanboy. I deal with facts. I like AMD, and i would still have 2x Fury X in my rig if it werent for the huge struggle with drivers and stuttering and crossfire problems… now i just have a single Titan X, no stutter, no sli problems, just smoothness, and now im waiting for someone to release something thats massively better.