Nvidia GeForce RTX 2080 50 per cent faster than GTX 1080

[kompukare]

Far too many paid mouthpiece 'journalists' around these days.
Wonder how many corporate PR teams asked PCW back in the day to censor Guy Kewney, Computer Shopper's MacBiter? And hopefully back then the editors told the PR teams to get lost.

[Corky34]

Apparently I heard some of the touted features which are not working in Vega now are broken(or won't be supported),so its hard to say,but at least with FP16 Vega probably will be longer lived than Pascal IMHO.

AFAIK they all have proper async compute, that is they can all perform calculation that are not dependent on another calculation being done first, and they can (afaik) all do double precision arithmetic (FP64). Where things differ is in how they process the calculations, if the 'cores' (they're more like ASIC's but whatever) are either designed to handle 8, 16, 32, or 64 bits per clock cycle and it's how they go about handling workloads that fall outside of what they were designed for where the differences lie, rapid packed maths allows two identical 16 bit calculations to be processed in a single 32 bit pipeline instead of taking up two separate ones, on the other hand using nothing but 64 bit 'cores' consumes more silicon, power, memory, etc, etc, so ideally you want to match the 'cores' width with the size of the calculations you intend to perform.

I'm only saying that because i recently got myself proper confused talking about that on another forum, knowing me I've probably got it all wrong again but hey ho.

[CAT-THE-FIFTH]

Wut:

https://www.hardocp.com/article/2018..._distribution/

I had mentioned this last week in our forums, but I have gotten a little more information about the amount of control that NVIDIA is exerting around its RTX 2080 and 2080 Ti card launch.

First and foremost, NVIDIA has demanded that its AIBs tell NVIDIA who will be reviewing the AIB's custom RTX 2080 and 2080 Ti cards. We were forwarded emails from other reviewers, from the AIBs that were asking specifically, at NVIDIA's direction, "Who will be performing the review content?" "What is that person's phone number and email address?" That is a bit odd, as we have never seen this before in 20 years of reviewing video cards. AIBs in the past have been left to pretty much operate their own review campaigns on new video cards, but that seems to have come to an end. From these lists of reviewers submitted to NVIDIA by the AIBs, NVIDIA has put together its own list of "approved reviewers," and sent their approved list back to the AIBs in order to let them know who they are allowed to sample review cards to. Much like NVIDIA exerted control over AIB's and OEM's brands with GPP, it is now exerting control over who the AIB has review its own custom cards.

This is where it gets a bit more interesting, and likely should give you concern with any leaked benchmarks you see on the web. NVIDIA is not allowing its AIBs to distribute drivers with their review cards. For a reviewer to have access, he must first sign NVIDIA's multi-year NDA (which is fine if you are "just" a card reviewer), then he will log into a protected site which is most likely a secured version of GeForce Experience in order to obtain the driver, and download from there into a specific machine with the new RTX card being present. If you are seeing any benchmarks between now and the ~20th (we think the 2080 launch and 2080 Ti launch will be split on different days possibly), you are likely not seeing cards benchmarked with its launch driver. So keep that in the back of your mind as you see performance leaks come forward.

It also means all performance leaks might not be that accurate.

[Safetytrousers]

Wut:
It also means all performance leaks might not be that accurate.

'If you are seeing any benchmarks between now and the ~20th, you are likely not seeing cards benchmarked with its launch driver.'
That seems like good news then, because the launch performance may be even better than the leaks.

[kalniel]

Drivers nearly always quickly improve in the early days anyway, nothing new there.

Locked software for reviews is nothing new either, pretty standard practise. However the 'approved list' of reviewers is a crap move if true. I'd be generating random names if I were a review site and/or be telling nVidia that all reviews are carried out by a staff team with no single individual solely responsible.

[Biscuit]

The amount of control Nvidia is exerting is pretty suspicious if you ask me.

So much evidence that pre-ordering this cards is a terrible idea... just wait until the independent reviews are out....

[XeNoNF50]

Think the issue here is more with the approved list. Everything else seems standard, right?

[peterb]

The amount of control Nvidia is exerting is pretty suspicious if you ask me.

So much evidence that pre-ordering this cards is a terrible idea... just wait until the independent reviews are out....

I don't understand this rush to pre-order - it hasn't made existing cards obsolete - but I guess its the same motivation that drives people to buy the latest smartphone in the hours before/after release.

[Safetytrousers]

Yeah I don't think I've seen any comments saying they'll be 'slow' (they shouldn't be at the price)

Most of what I've seen is people questioning the performance. I've seen just one comment on all the net about how someone would not buy at these prices no mater how good the performance was.
Have you not noticed all the comments about shady marketing and that it seems NVIDIA have something to hide?

[kalniel]

Most of what I've seen is people questioning the performance. I've seen just one comment on all the net about how someone would not buy at these prices no mater how good the performance was.

I know you've just joined Hexus, but a quick look here shows many people are questioning the price, not the performance.

[CAT-THE-FIFTH]

Regarding that TH article there seems to be many issues at TH,to the extent the editor of the German version,actually rebuked it:

https://www.tweaktown.com/news/62951...lic/index.html

What the hell happened? Tom's Hardware has changed in the Avram Piltch era. In recent months, some veteran staff have jumped ship and many others are desperately looking for a life raft before the full implementation of Tom's Hardware 2.0, a Google SEO driven site. These sites, like Tom's Guide, don't break new stories but rather identify trending topics and then quickly build content to match. Purch has an internal team scrubbing the internet each day that assigns stories to different teams inside the company. The Tom's Hardware 2.0 formula has not sat well with veteran journalists and editors that continued the twenty-two year legacy of Tom Pabst.

The first forward-facing look at the current internal struggles come from Igor Wallossek, editor-in chief of Tomshardware.de, a licensed spin-off of Tom's Hardware Germany. Igor, a 10-year veteran of Tom's Hardware, writes in his German response to the most recent controversy: "With this attitude, the interim editor-in-chief also contradicts my very personal work, since I am directly involved in the launch article. Who is to believe the hard-won results after so much advertising balsam? And hello? testing complex components requires more basic knowledge and expertise than the superficial clowning of notebooks and colorful Lego heaps.." - Translated.

The statement refers to Avram Piltich's editorial writing for Laptop Magazine, Tom's Guide, and Space.com where he wrote about mainstream topics like notebooks and Legos. Avram has never authored content that meets the high expectations of a Tom's Hardware journalist much less qualified to be the editor-in-chief under the former format. When Avram came to Tom's Hardware to transform it into a larger Tom's Guide filled with clickbait content, he started by writing the topics he understands, Legos and non-technical notebook editorials. The new direction and content didn't go over as well as similar articles at Tom's Guide and Space.com, both also owned by Purch - the owners of Tom's Hardware and Anandtech. Avram Piltch has doubled down on statements via his GeekinChief Twitter account, and Igor went nuclear in the Tom's Hardware forums with more claims of Avram's inexperience and even a breach of contract related to the German site's licensing. Both comments were scrubbed by forum moderators. Later in the day we had our second look at infighting at the media company. An ethical but disgruntled moderator enabled the posts again for public viewing before issuing a public statement and later resigning his position. We can only imagine what the C-Suite will be like in the Purch offices on Friday. We hear Future plc, the publishing company in the middle of acquiring Purch, comes to the office for the first time at the end of the week. In the press release announcing the acquisition, Future placed a strong emphasis on the reputation and technical detail from the flagship property, Tom's Hardware. The current leadership inside the company clearly no longer cares about the site's reputation and doesn't have the technical knowhow to publish highly technical content outside of a few writers that have remained only to to see the rubbishrubbishrubbishrubbish storm pass.

It seems the German version of the website is run by different people who thought differently about the article.

Also here is what the founder of Tom's Hardware has said:

https://www.tweaktown.com/news/62960...dal/index.html

Thomas Pabst, founded of Tom's Hardware, spoke out publicly for the first time about the RTX "Just Buy It" article that ran last Friday on the site he founded. Tom has not been vocal about Tom's Hardware so this was unexpected : "In regards to the RTX scandal at Tom's Hardware, Pabst said: "You asked 'What happened, Tom? What has become of you?' Well, there's only one person who can answer that! Tom has become a daddy of two boys and doesn't have anything to do with Tom's Hardware anymore. So far so good". Pabst continued, pondering: "What does Tom think of the article you love so emphatically?" "Well, I'd say Tom would have been less kind than you have been with his assessment! It is ultimately ridiculous, it is indeed suicidal, and its conclusions are epically nonsensical. There is value in being an early adopter? Aren't we, who are impatiently into the latest tech, sorely aware of what we keep doing to ourselves when we purchase technology at high prices 'ahead of the curve'? It's maso-f***ing-chism!" "We make ourselves paying beta testers, and wait for software (and often other hardware) that will hopefully bless our brand new tech with the meaning and usefulness they simply do not have by the time of our premature purchase! There are no RTX games, but there is value in adopting Turing early, because current games might be faster than on our 1080ti SLI setup? Yes, this is madness, and good old Tom is scratching his head no less than you are, Steve! There's got to be value in masochism!"

[Corky34]

Tom's Hardware was added to my search engine blacklist ages ago, i think it was because Tom's Guide or inane forum posts kept filling up my results no matter what i searched for.

[Biscuit]

Tom's Hardware was added to my search engine blacklist ages ago, i think it was because Tom's Guide or inane forum posts kept filling up my results no matter what i searched for.

I never thought they were that bad, until I read that article... which I genuinely thought was satire.

[Corky34]

It wasn't the main site (is Tom's Hardware still the main site?) that made me block them, the main site produces some very well written in-depth articles that I'd definitely read, unfortunately it got caught up in the blocking of Tom's Guide and the forum posts.

[Corky34]

Wut:

https://www.hardocp.com/article/2018..._distribution/

It also means all performance leaks might not be that accurate.

I make no claims of validity but thought I'd throw another point of view out there...

That's just a big can of nonsense (and I initially wrote another word there). NVIDIA always has tracked what media gets what AIB samples, period. You know who does that as well? AMD, they even regulate what brand and sample end up at what reviewer. How conveniently he forgets to mention that.

[XeNoNF50]

I make no claims of validity but thought I'd throw another point of view out there...

I mean most of what he wrote did seem standard practice, though the content in that hardocp piece may shock people who don't know otherwise.