The FE has a really nice cooler, it boosts like crazy to 1980 but it will then thermal throttle down-clock after a while. You really want to keep the FE at stock because its at the thermal limit of the cooler out of the box.
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
1080ti Custom vs 2070S FE vs 5700XT 50th AE Bench Thread
- Thread starter Besty
- Start date
The FE has a really nice cooler, it boosts like crazy to 1980 but it will then thermal throttle down-clock after a while. You really want to keep the FE at stock because its at the thermal limit of the cooler out of the box.
My 2070S settles at 2070MHz and doesn't throttle, only running at 65-70. Ok it does throttle but only to 2050MHz where it's solid.
Associate
- Joined
- 23 Oct 2013
- Posts
- 1,253
It's good to see these cards compared on the same system.
Can you also try 3440x1440 res as more and more are moving to ultrawide? (would supersampling enable benchmarking a res higher than monitor res, if you don't have ultrawide?)
dumb question, but what's the point of the 5700XT set to match the 2070S boost clock?
I can understand stock benchmark and also oc'd values. But, why try to set it to act like the other card (if that's what is happening here)? Would this ever happen in reality? Wouldn't people just buy a 5700XT and either keep it stock, or OC (without trying to match a different card)?
Can you also try 3440x1440 res as more and more are moving to ultrawide? (would supersampling enable benchmarking a res higher than monitor res, if you don't have ultrawide?)
dumb question, but what's the point of the 5700XT set to match the 2070S boost clock?
I can understand stock benchmark and also oc'd values. But, why try to set it to act like the other card (if that's what is happening here)? Would this ever happen in reality? Wouldn't people just buy a 5700XT and either keep it stock, or OC (without trying to match a different card)?
re: 3440x1440 - I will give it a go next week once I have my block on, I have work commitments today.
The issue regarding clocks is quite simple, what Nvidia have done here is placed a very good cooler onto the 2070S FE and allowed it to boost into a thermal wall around 1907-1925mhz whilst keeping the voltages low. The chip is a very refined design and the card is tuned to within an inch of its life out of the box.
The 2070S FE design is very similar to the 1080 Classified I bought years back.
AMD have yield issues with their chips and thus the default voltage is set too high on the 5700XT, combined with the choice to give their partners a treat (i.e. make some money on this launch) by going with the blower means that the 5700XT is pretty crippled out of the box. Once you set the card up properly (reduce volts and increase the clock) it can easily match the 2070S FE boost. I think this is a fairer test although I have included stock results too.
I don't think its particularly fair to compare the cards without matching their boost clock speeds because the 3rd party 5700XT cards will resolve this issue.
Associate
- Joined
- 23 Oct 2013
- Posts
- 1,253
^ I understand that completely, I just don't understand why it's set to the 2070S boost clock. If you feel the 5700XT can be oc'd safely further, then why not do that? Of, if the existing OC (to match 2070S) is an unrealistic long-term OC then, also it's not really viable. I hope I'm explaining myself well enough. Sure, check OC'd 5700XT (which should better represent AIB cards), but it should be an OC that suits the card, as a reliable/stable realistic OC, not something bound to a different card, designed to a different architecture.
Stock 5700XT FE vs OC'd 5700XT (vs stock 2070S and OC'd 2070S) - all oc-ing should be realistic oc's to suit their own designs, with values that should be achievable by all users (without special hardware)?
Or am i still missing the point?
Stock 5700XT FE vs OC'd 5700XT (vs stock 2070S and OC'd 2070S) - all oc-ing should be realistic oc's to suit their own designs, with values that should be achievable by all users (without special hardware)?
Or am i still missing the point?
which XTs and which 2070S's are we comparing to in this comparison?
I see a variation in XT pricing
The standard cheapest XT @ £380, and any 2070S. There are no other XT there so the cheapest is the same as the most expensive. (FYI Powercolor announced that their custom 5700XT will go for $399).
Only the FE is at MSRP of £475 while even the blower cooler Palit is more expensive, and only 1 AIB card (2+ fan) is sub £500, the KFA one at OCUK.
All the rest of 2070S in UK are at £530 or more, with Asus over £600.
£380 to £530 is 40% price increase, and I do not count the Sapphire at £375 OCUK has today.
@Besty have you posted somewhere image of your 5700XT curve? Couldn't find it, only the final point.
Asking because I found that we can undervolt the whole curve not only the final top point. It allows the card to have different power states and speeds in between.
Asking because I found that we can undervolt the whole curve not only the final top point. It allows the card to have different power states and speeds in between.
A quick google showed a few retailers selling XTs at around £430. I don't think its fair to take the chapest price from one brand, and then then most expensive from another and compare.
You're not wrong in that there is a significant price difference, but its really not £200.
Given that I have to reply to you properly and you diverge the discussion in this place, I will reply on the Navi discussion.
Now in process of add the Division 2 scores and i found the problems (over-heating DDR4 ram 1.6vdimm) with watchdogs 2 and will re-run for 5700XT.
2070S FE and the 1080ti HOF are virtually identical in D2. I got much more heat soak in this game on the 5700XT with fan on auto. I will see what clocks I can get versus fan settings tomorrow.
I am confident that the 5700XT will put in a significantly improved showing once the heat issue is resolved.
None of these cards hit 90fps in 1440p, I am lucky enough to have a monitor where variable rate refresh works on all three cards but i would sooner play in 1080p because the frame times are significantly lower on all three cards.
A 6-core CPU will help in this game but we really need a 2080ti to get 1440p at Ultra at decent performance levels.
2070S FE and the 1080ti HOF are virtually identical in D2. I got much more heat soak in this game on the 5700XT with fan on auto. I will see what clocks I can get versus fan settings tomorrow.
I am confident that the 5700XT will put in a significantly improved showing once the heat issue is resolved.
None of these cards hit 90fps in 1440p, I am lucky enough to have a monitor where variable rate refresh works on all three cards but i would sooner play in 1080p because the frame times are significantly lower on all three cards.
A 6-core CPU will help in this game but we really need a 2080ti to get 1440p at Ultra at decent performance levels.
Last edited:
@Besty if I remember correctly, you got your waterblock yes? Have you had look at the More Power Tool for the Navi cards?
Apparently can hit successfully 2300, and change the memory timings. It also allows 1225mv not just 1200mv as it was up to now.
https://forums.overclockers.co.uk/posts/32925080/
Apparently can hit successfully 2300, and change the memory timings. It also allows 1225mv not just 1200mv as it was up to now.
https://forums.overclockers.co.uk/posts/32925080/
block testing will start thursday
the voltage slider my AE goes up to 1.35v with powertable softmod muhahaha
block testing will start thursday
the voltage slider my AE goes up to 1.35v with powertable softmod muhahaha
Yeah I saw saw. But don't want to touch it while with the ref cooler atm.... not saying that I haven't had a look at it
Yeah I saw saw. But don't want to touch it while with the ref cooler atm.... not saying that I haven't had a look at it
lol yeah it’s tempting but i think that setting is on the degradation limit of the silicon and we need to keep temps well in check before trying it.
i personally think it will be a re-run of the 1080 i.e we will need a big power limit and voltage is not the problem.
Water Testing results for the 5700XT have been delayed until next week as i had a hardware fault with my work laptop which took priority.
The EK block is fitted and running on the 5700XT , impressions are that air-max clocks are the target with the benefits of no clock drop over the hour mark due to heat soak and near silence.
The EK block is fitted and running on the 5700XT , impressions are that air-max clocks are the target with the benefits of no clock drop over the hour mark due to heat soak and near silence.
So would you say that most of thehe potential is already being realised on the air cooled cards? Sounds like water cooling on the big Navi cards could be beneficial.
Associate
- Joined
- 17 Jul 2019
- Posts
- 91
According to Hardware Unboxed the delta in clockspeed on water was negligible. The biggest improvements comes from greatly reduced temps and noise. At which point the cost becomes so big that you're better off getting a AIB 5700xt (or the 2070s).