LOLYou mean I cant use it with my 166 Mhz Pentium???
Both companies wanted to achieve the same thing: introduce some cost reduction for the smaller memory version to increase margins. Nvidia was smarter about it by making sure that the cost savings would end up in their pockets in the form of yield increases.
I would rather say, they were in the position to play that card. AMD OTOH also has to find means and methods to get their marketshare up, which usually involves comparatively attractive cut down models that eat into the margins.
Dosen't RX470 do the same thing for AMD?
kinda but not really, it creates internal competition for the rx480 4gb so people who really want the rx480 won't get the rx470 and just wait until they can't get the rx480 4gb then they get the rx470 as an alternative (if its available) or another board.
Pricing is just screwed up....
Pricing is just screwed up....
Yes, except that the name of their cut down part explicitly conveys the message that it's lower performance.
Does it really make sense to have 3 SKUs that are so close to each other both in price and perf? Maybe it does...
D
Deleted member 2197
Guest
Interesting undervolting test with Palit GTX 1080 Jetstream -- trying to extract highest clock rate and run card as efficiently as possible by lowering the chip voltage.
They were able to reach 1759 MHz using only .8v amounting to power consumption of approx. 130 watts. On the flip side they achieved a cpu clock or 2000 MHz using .95v with power consumption roughly at 187 watts.
Original link:
http://www.forum-3dcenter.org/vbulletin/showthread.php?t=574892
They were able to reach 1759 MHz using only .8v amounting to power consumption of approx. 130 watts. On the flip side they achieved a cpu clock or 2000 MHz using .95v with power consumption roughly at 187 watts.
Original link:
http://www.forum-3dcenter.org/vbulletin/showthread.php?t=574892
D
Deleted member 2197
Guest
Nvidia highlights Intel Deep Learning benchmark mistakes
http://hexus.net/tech/news/industry/95893-nvidia-highlights-intel-deep-learning-benchmark-mistakes/Intel claims:
- Xeon Phi is 2.3x faster in training than GPUs
- Xeon Phi offers 38% better scaling that GPUs across nodes
- Xeon Phi delivers strong scaling to 128 nodes while GPUs do not
It seems that these claims have hit a sore spot with rival chipmaker Nvidia. In a blog post earlier this week the green team was keen to point out that "newcomer" Intel was using out of date benchmarks. Nvidia says that deep learning is such a fast moving field some companies might not be able to keep up with cutting edge developments. A case in point is that Nvidia's own architecture and software "have improved neural network training time by over 10x in a year by moving from Kepler to Maxwell to today’s latest Pascal-based systems".
...
Offering some constructive criticism for Intel, Nvidia thinks that "deep learning testing against old Kepler GPUs and outdated software versions are mistakes that are easily fixed in order to keep the industry up to date". Nvidia's blog post concludes that while it is "great that Intel is now working on deep learning," the company should endeavour to "get their facts straight".
http://videocardz.com/63252/nvidia-launches-geforce-gtx-1060-3gb
And the 3gb version of the 1060 is launched..... just don't think 3gb is enough for this card....
And the 3gb version of the 1060 is launched..... just don't think 3gb is enough for this card....
First official notification of 1060 3GB we got was from Inno3d, followed by EVGA:
http://www.inno3d.com/products.php?refid=1&subid=18
http://eu.evga.com/articles/01047/evga-geforce-gtx-1060-3gb/
So it's official now: 3 GB and 1152 ALUs, with only the „3GB“ differentiating the card from the 6-GB-SKU on the outside, confusion inbound.
Pricing for germany seems to be 219 EUR including all necessary taxes.
http://www.inno3d.com/products.php?refid=1&subid=18
http://eu.evga.com/articles/01047/evga-geforce-gtx-1060-3gb/
So it's official now: 3 GB and 1152 ALUs, with only the „3GB“ differentiating the card from the 6-GB-SKU on the outside, confusion inbound.
Pricing for germany seems to be 219 EUR including all necessary taxes.
Ike Turner
Veteran
I dream/hope that the rumored Surface Book "end of the year tweaked/updated version" will feature the newest Pascal notebook GPUs...
A1xLLcqAgt0qc2RyMz0y
Veteran
So I take it you are NOT confused with the AMD RX 480 8GB and the RX 480 4GB.
The only difference between the two is memory, one is 4GB/7Gbps, one 8GB/8Gbps, the GPUs are identical. GTX 1060 3GB has less CUDA-cores and texture units in addition to different memory amount.
Because you was confused seeing a 290 4B and 8Gb configuration ?
The small ( and i reallly say small ) problem is that the 1060 3GB have a different configuration that the 1060.. something more in the line of the 970 and 980 . Peoples are used to see differentt memory config on a same gpu name .. different config of the core ( and so performance ).. thats a bit different..
Last edited:
A1xLLcqAgt0qc2RyMz0y
Veteran
So "Storm in a teacup" then.
Seems like people are jumping to conclusions about performance of the GTX 1060 3GB vs the 6GB model yet have no problem with the RX 480 4GB having slower memory speed vs the RX 480 8GB.
Double standard.
We have a different memory configurations ( understand size ) on GPU's , since a good decade.. a same gpu name for different numbers of cores and textures units, mean a different performance . need new habitude maybe.... Nvidia is really good, as Apple to bring new habitude for consumers. Its called marketing.. This card should been callled 1050TI... or whatever, but well... who cares, you willl buy a 1060 no, and certainly know that this card have less cuda cores, less textures units that the 1060, its indicated by the size of the memory ?
I joke, but seriously i dont care about it, if someone is enough stupid to buy a 1060 3gb thinking he just buy a 1060 with less Vram, i find that funny. Somwewhere he deserve it ..
On an other way, this card will have excellent performance for his price, making it a good, an excellent deal. So, all good .
Last edited:
No double standard. Officially AMD specs are 7 Gbps memory for the RX480 with 8Gbps at the discretion of partners. Either ways..it's far better than NV's approach where the GPU itself is cut down and yet branded the same.
Tbh I'm actually a bit surprised this was cleared by their legal team after the 970 class action. IMHO this would have been a good opportunity to resurrect the GTS branding and name it 1060 GTS.
PS: With a TMU to ROP ratio of 1.5:1, I think its probably the lowest we've seen in a few years now.
Actually not true, 7 Gbps is AMD spec only for 4GB models, 8GB models AMD spec is 8 Gbps
We've already discussed this in another thread. From AMD's website - http://www.amd.com/en-us/products/graphics/radeon-rx-series/radeon-rx-480
MEMORY CLOCK SPEED (MHZ) - 1750 or higher
MEMORY BANDWIDTH - 224 GB/s or higher
Regardless of what they say there, on launch they were really clear about it, 4GB / 224 GB/s and 8GB / 256GBs, with the possibility of AIBs bringing different memory clocks on both products
