Is it possible that the RX 6800XT is faster than the 3090? - Page 4 - Video Cards - HWzone Forums
Skip to content
  • Create an account
  • About Us

    Hello Guest!

     
    Please note - in order to participate in our community, comment and open new discussions, you must join as a registered member.

    Our members enjoy many advantages, including the ability to participate in discussions, enjoy raffles and promotions for members of the site, and receive our weekly content directly by email.

    Do not like being harassed by email? You can register for the site but do not submit your registration to the weekly email updates.

Is it possible that the RX 6800XT is faster than the 3090?


Recommended Posts

  • תגובות 297
  • Opens on
  • Last comment

Prominent participants in the discussion

Prominent participants in the discussion

Popular responses

What closes with these messages? What do they contribute? Porn of box card boxes? Not so clear to me ...

If so, at a good time we finished from the first end where we measure the eight titles we originally tested on 6800xt so we have a reference for reference and comparison against each other. We are very happy that we insisted on finding a good 3080 card, and were not satisfied

Do not understand this narrative of "Fine Wine". I do not know in which communities you hang out, in the ones I hang out in it is mostly such a network joke. You could call it "drivers that get better with time" ... you could also call it bad drivers

Attached Images

Thanks for the caveman input. It's also something I found in a web search - just like you wrote.

Since lanzar did timespy at least that, and we uploaded results, you can go over them.

An improvement of only 10% compared to Stock was achieved, due to the limitation of the current limiter in the card (limited to 320W) which prevents

Unfortunately, work above 2.5ghz - this despite the card's ability to give up to 2750MHz. 

 

** Next we will try to put a BIOS where the limit is removed (BIOS from Custom) and see where it is possible to go up. A rough estimate would be that if achieved

Another 200mhz, this is an improvement of another 5-7% to about a rough order of magnitude.

 

Since in the architecture of the RDNA2 frequency A cache locked to the core frequency of the GPU, the resulting performance improvement is quite

Linear to running frequency. Therefore perform an interpolation to improve performance in his individual case, relatively works nicely we have seen.

Get a linear improvement to the running frequency almost completely 1: 1.

 

Almost, as is well known, there are additional inhibitions in a computer system that come from the platform itself (processor, memory, ,

BUS, and so on ...) that prevent the full 100% from materializing.

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites

And friends, this is a forum that was once very very active, such and other tests were welcomed and attracted more people who gave their own results, and even then there were tests across the network, some more reliable and some less.

You will receive everything in a good spirit. It's totally for everyone.

Post Reply Direct Link To This Post
Share on other sites

Ok, lanzar found that we probably ran the metro yesterday in slightly different amusing settings for the light:

Lior performed without advanced physiscs and without hair works, while lanzar ran with (because these are some of the ultra-default settings).

lanzar is currently aligning with the settings that Lior has run, and brings us an updated measurement. Stay tuned ...

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites
Quote of KobyC92

It seems to me worthwhile and worth noting to anyone reading this thread. Performance is under OC on a single card. We cannot know if it allows for an average interval, well above average or at all below. The reason you check for a stock configuration is this. That could be a card from the same series that would perform almost no OC at all. I personally would not base a purchase based on OC performance. 

 

Excellent question Kobe.

From what we were able to check across the network, it mainly depends on the card model you are purchasing. There are two individual models that I went through

On them that always do 2.7hz. These are the first two cards I mentioned at the beginning of the thread, the In the version

His premium, and the Merc 319. In the present case the lanzar works with a phantom version.

 

These are top edge custom cards, priced at $ 750 instead of $ 650 MSRP, they are huge, 3 slots, length 34 plus cm, 3

Fans with a large adsorption, and weighing almost 2 kg. Explanation that these are versions that probably get chips that are binned,

When the chips are of lesser quality Assigns to the cheaper reference versions that cost $ 650.

 

I mean, that when we talk about For the 2.7Ghz dialing area, we explicitly mean cards from these models

Specifically known to work at these frequencies. Also the rush is simpler than you think, it's all a slider moving frequency up

The end is right and that's it. There is no control over the core voltage, and no more settings to make - everything is locked. It's just shifted to the right for 3 seconds of work

And close the window. How simple and impressive.

 

Regarding the sample size, lanzar is now ordering two more such dance cards, and when they arrive they will check their working frequency as well, we will report

Here neatly. So the individual sample we have will later go up to 3 tickets.

By the way, anyone who knows lanzar personally and closely knows, that he assembles systems. Goes through a considerable amount of hardware including

Premium hardware. I have no doubt that in the next quarter or two the sample will increase even more, since as a component of systems for its customers, it takes care

Choose for them very carefully the products, the quality of workmanship of the assembly (with Water) which is To them, and calibrate

And configure the machine at the highest level possible. When he measures, it's always the top that can be achieved, because he does everything by the

book. Is one of the most experienced I have known in this field in the country. In fact, that's how I met him a few years ago, just by chance.

After that I found out he is one of the guys from the forum as well. small world : )

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites
Quote of nec_000

Ok, lanzar found that we probably ran the metro yesterday in slightly different amusing settings for the light:

Lior performed without advanced physiscs and without hair works, while lanzar ran with (because these are some of the ultra-default settings).

lanzar is currently aligning with the settings that Lior has run, and brings us an updated measurement. Stay tuned ...

 

Ok, the corrected run was performed:

Indeed an improvement concept to 118.4FPS (and now also the benefit of Disappeared we checked again also its name on 10900K). 

Accordingly, we updated the findings on the previous page.

 

Summary:

In the metro title an improvement is achieved from 105FPS in stock, to 118.4FPS 

9.thumb.png.54d52aac06c592ea34a3e20879011d3c.png

 

8.thumb.jpg.8c42ab287e766c52cd9525b193747438.jpg

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites

The caveman,

 

Starting from 1440P because this is the screen that lanzar has. ** We will connect later And will also measure . patience.

 

However, as long as we do not have CPU limitations that might interfere, then we do not expect a linear change

Significant (or at all) in the transition between resolutions. That is, we expect an improvement of size X% obtained at 1440P,

There will be the same percentage improvement from size X, plus minus, also in the other resolutions Or 1080P for that matter, and as long as the processor
There will be no element that of course limits low resolutions.

 

This is because the RDNA2 architecture works against the cache and not the VRAM. Hence the transition between resolutions

Less dominant than before.

 

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites
Quote of Lanzar

And friends, this is a forum that was once very very active, such and other tests were welcomed and attracted more people who gave their own results, and even then there were tests across the network, some more reliable and some less.

You will receive everything in a good spirit. It's totally for everyone.

The tests are welcome and thank you for doing them, the bombastic statements of other people here and the clickbait title are the problem. 

Also the attempt to show oc as a feature even though every person with a drop of experience in the field knows that oc is first and foremost a gamble and it is never possible to look at what one examiner has received and from there draw sweeping conclusions. 

Post Reply Direct Link To This Post
Share on other sites

Clickbate headlines are meant to attract attention, with the goal of definitely getting people inside. 

That if women a dull and grayish headline, the exposure on the subject would be accordingly, low, only a handful who happened to come in to see OC would have entered.

We seek to produce maximum possible exposure. This is totally for the benefit of the members and the whole community, the forum is nothing but a tool to enable

The desired exposure. And the title is one means in the arsenal of tools to produce the exposure.

 

** Meanwhile lanzar's back is broken, his wife is about to throw him out of the house because he has suffered from 4K TVs moving from living room to room ...

A little respect to a friend who is sweating for all of us, I asked him to focus on 4K as soon as possible. which man. Patience friends ...

It really comes down to choosing the right appreciation for working hours for us, it's from the heart and complete volunteering of course.

 

To the subject ,

At this point we expect, the so-called preliminary assumption, that if we saw an improvement of size X at 1440P, say 13% for example,

So in the same title, also in It seems that the size of X percent is the same. This is required to happen, except for slight deviations here or there

Due to symbiosis with the CPU / motherboard / platform work, which may tilt slightly here or there.

 

By the way, where 1440P has reached the CPU capability ceiling, beyond Will be able to show us a better improvement

Following the same OC, since then the CPU will not interfere as much as it interfered with 1440P.

* In the last headline we have probably seen such a case, and we expect that- We will get a better percentage improvement because of that.

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites
Quote of Lanzar

And friends, this is a forum that was once very very active, such and other tests were welcomed and attracted more people who gave their own results, and even then there were tests across the network, some more reliable and some less.

You will receive everything in a good spirit. It's totally for everyone.

Very happy and interested. The truth is that the missing estimate point is the same test you have without OC. To make sure that the addition to the performance is not due to a new driver, etc. 

Post Reply Direct Link To This Post
Share on other sites

Yoav,

From there we actually started. This is what Lanzar first examined, when he encountered gaps between his findings and other reviews.

In a joint consultation that we talked about during the weekend, even when we were investigating the issue, we realized that a reference must first be measured,

Understand whether the gap is only from the running frequency, or reasons foreign to it. We went through everything, the infrastructure specifications of ,

Driver, firmware, and so on. One of the things we came up with is a firmware update to his motherboard for example. Made problems before,

did not work timespy. We saw meaning for dual rank memory and the like ...

So we asked to make a frequency reduction to that of the reference model, to get a starting point for reference, understand where to start and compare

For measurements of others on the network. We compared Lior's findings and obtained the same (with an acceptable standard deviation). It was the anchor to go out

On the way. In fact we did That Lior's work is accurate And the reprehensible findings.

We then examined while isolating a parameter, quickly alone what it does, Separately what does, together both what they do,

Attempt to recover findings from other speeders who posted results online with the same frequency and / or the same card, and the whole business connected to

The insight we share with you about the importance of not doing OC to VRAM and that there are situations / cases where Risen 5000

Provides performance improvement that is more powerful than the corresponding Intel and / or because of the new technology of direct access to the memory of the CPU.

We arrived at this thread cooked already after several days of research and we are still learning while on the move of course and performing 

reassessment of our assumptions as discoveries emerge from the field and tests.

We will continue to strengthen the learning and if we get 3090 or 3080 in the near future, or if he helps us with friends and measures for us with him, we will ...

 

Edited By nec_000
Post Reply Direct Link To This Post
Share on other sites

Join the discussion

You can then join the discussion and then join our community. If you already have an account with us, please Log in now To comment under your username.
Note that: The comment will appear to the surfers after approval by the board management team.

guest
Add a comment

×   The content pasted is with formatting.   Remove formatting

  Only 75 emoji are allowed.

×   Your link has been automatically assimilated.   Show as regular link

×   Your previous content has been automatically restored.   Clear all

×   You can not paste images directly. Upload or insert images from URL.


×
  • Create new ...

At the top of the news:

In June everything will work out? The MWC 2021 exhibition begins with a Corona-adapted version

Instead of a grandiose mobile celebration in Barcelona - we get a relatively small event in Shanghai this week, and plans for a European comeback this coming June The MWC (Mobile World Congress) was the first notable technological event to suffer last year's coronation, with naive restrictions in the first stage, Complete cancellation of the physical event afterwards - and at the same time promises that next year everything will return to normal, when it is likely that at that point in time no one guessed or predicted that even in 2021 ...

new on the site