2070 Oc Scanner

[Fatic]

Hey guys i have gigabyte rtx 2070 i will start oc scanner on the msi afterburner.i see top clock 2090Mhz but i cant manuel oc this value.How can i Using 2000mhz on the stable.

 

[Vayra86]

Hey guys i have gigabyte rtx 2070 i will start oc scanner on the msi afterburner.i see top clock 2090Mhz but i cant manuel oc this value.How can i Using 2000mhz on the stable.

You cannot. GPU Boost 3.0 > will automatically adjust your clocks based on temperature (and load). Turing and Pascal GPUs have a fine grained temperature control that you cannot adjust, it works by dropping 'boost bins' at every 5 C interval. So the higher your card temps, the more boost bins (13 mhz each) you lose while gaming.

You have a good OC with a 2090 mhz peak clock, I'm sure at full load it will level out at around 2000 or more. Which is true for most cards in the Pascal and Turing line up.

Note, even while OC scanner says you max at 2090, you can probably still try pushing the core clock a bit higher. As long as you're stable, it just means you will also end up a bit higher under load.

 

[Fatic]

You cannot. GPU Boost 3.0 > will automatically adjust your clocks based on temperature (and load). Turing and Pascal GPUs have a fine grained temperature control that you cannot adjust, it works by dropping 'boost bins' at every 5 C interval. So the higher your card temps, the more boost bins (13 mhz each) you lose while gaming.

You have a good OC with a 2090 mhz peak clock, I'm sure at full load it will level out at around 2000 or more. Which is true for most cards in the Pascal and Turing line up.

Note, even while OC scanner says you max at 2090, you can probably still try pushing the core clock a bit higher. As long as you're stable, it just means you will also end up a bit higher under load.

Thx for reply.1900mhz is stable on my card?What is your idea?Because i have 4k monitor.

 

[EarthDog]

You need a better card for 4K. OC will only improve things a few %. It isn't a change from unplayable to playable.

 

[Vayra86]

Thx for reply.1900mhz is stable on my card?What is your idea?Because i have 4k monitor.

We cannot tell you what is stable and what is not, every chip is a bit different in how well it clocks & responds to higher temperatures/voltages.

It is a good custom 'to find your max OC' and then dial down the core clock a little bit to stay safe, with GPUs. A good 'quick' test for stability is Unigine Valley and/or Heaven, and in that order. Try settings in Valley, push it to the max until you crash it, dial it down a bit, and then run both Valley & Heaven. Heaven is a bit heavier on the GPU (adds heavy tesselation). If it survives Heaven, run 3DMark Firestrike. If all three are OK, you've got a pretty stable setup.

Regardless like @EarthDog said, for 4K the amount of overclocking will not change things from unplayable to playable. Its better to tweak in-game settings a bit for best results.

 

[EarthDog]

Heaven is a bit heavier on the GPU.

Really?

Ironically, in their description, Heaven says this is GPU bound... and it may have been in 2009 when it was released. But today, its a CPU bound benchmark in many cases. There are large gains to be had when spinning up your CPU in this benchmark (in particular using the HWbot version - I think in the consumer version you can adjust resolution, right?). Unlike UL/3DMark, the CPU isn't a variable in scoring so its all a horsepower thing, in particular on high-end cards. Valley is newer, but I am not sure it uses tessellation either which may be Heaven's only saving grace for stability testing (that and for older, DX9-11 games and Open GL 4.0 and earlier).

That said, I prefer to use the UL Benchmarks Time Spy or Fire Strike (Extreme) for testing and loop them... works a charm for me.

 

[Vayra86]

Really?

Ironically, in their description, Heaven says this is GPU bound... and it may have been in 2009 when it was released. But today, its a CPU bound benchmark in many cases. There are large gains to be had when spinning up your CPU in this benchmark (in particular using the HWbot version). Unlike UL/3DMark, the CPU isn't a variable in scoring so its all a horsepower thing, in particular on high-end cards. Valley is newer, but I am not sure it uses tesselation either which may be Heaven's only saving grace for stability testing (that and for older, DX9-11 games and Open GL 4.0 and earlier).

That said, I prefer to use the UL Benchmarks Time Spy or Fire Strike (Extreme) for testing and loop them... works a charm for me.

The guy runs 4K, mate. You try that, its going to put you royally below 60 FPS on a 2070 and/or at least royally out of CPU bound territory. The main reason for Valley is to quickly find a max to go on, and the reason for Valley is that 3DMark SysInfo crashes if you make clock changes and it has a very long startup time whereas in Valley you can do everything on the fly with the bench running.

I agree, for stability 3DMark is preferable, but not for finding max OC.

 

[EarthDog]

Yeah, I'm used to the Hwbot version with a wrapper.

Still, Heaven has seen its day, and it is past. It, like Valley, was last updated in 2013 (Heaven was created in 2009). There are better, more modern, tools for the job. But as always, to each their own.

NO sysinfo crashes here... and I just rebenched 4 RTX cards both stock and overclocked.

I agree, for stability 3DMark is preferable, but not for finding max OC.

How is this any different? Why would one use a program for 'max oc' then another for stability? That doesn't make any sense. YOu want stability in the max overclock unless you are benchmarking...but that isn't what we are talking about here either.

 

[Solaris17]

Really?

Ironically, in their description, Heaven says this is GPU bound... and it may have been in 2009 when it was released. But today, its a CPU bound benchmark in many cases. There are large gains to be had when spinning up your CPU in this benchmark (in particular using the HWbot version - I think in the consumer version you can adjust resolution, right?). Unlike UL/3DMark, the CPU isn't a variable in scoring so its all a horsepower thing, in particular on high-end cards. Valley is newer, but I am not sure it uses tessellation either which may be Heaven's only saving grace for stability testing (that and for older, DX9-11 games and Open GL 4.0 and earlier).

That said, I prefer to use the UL Benchmarks Time Spy or Fire Strike (Extreme) for testing and loop them... works a charm for me.

Agreed, Time spy passes a lot of the time but even normal fire strike will fail my titan when pushing the upper limit

 

[ppn]

2000 stable only if you set TDP to 250 watts somehow. my card is windforce3 only goes upto 200 watts, but the gaming and aourus versions surely can do more watts than 215. Simply put with 200 watts one can only get around 1800 at 0.95 volt. to hit 2000 you need a little bit more voltage and TDP headroom to avoid throttling.

 

[Vayra86]

Yeah, I'm used to the Hwbot version with a wrapper.

Still, Heaven has seen its day, and it is past. It, like Valley, was last updated in 2013 (Heaven was created in 2009). There are better, more modern, tools for the job. But as always, to each their own.

NO sysinfo crashes here... and I just rebenched 4 RTX cards both stock and overclocked.

How is this any different? Why would one use a program for 'max oc' then another for stability? That doesn't make any sense. YOu want stability in the max overclock unless you are benchmarking...but that isn't what we are talking about here either.

I just found it was quicker, for me as a lowly consumer working with regular 3DMark. I run Heaven, tweak clocks until it crashes, then work my way down until it doesn't. It saves me a crapload of time restarting a benchmark, loading screen, etc. 3DMark FSU is then just the stress test that I might run twice, maybe three times. But more often just once to find that its stable, like Heaven also was.

Another potential reason to run Heaven is because you can see how the card responds to high refresh/FPS, saves you a lot of tweaking in 3DMark settings.

 

[mstenholm]

Hey guys i have gigabyte rtx 2070 i will start oc scanner on the msi afterburner.i see top clock 2090Mhz but i cant manuel oc this value.How can i Using 2000mhz on the stable.

So I assume that you "fixed" your other Multi posts but no feed back from OP. I doubt that I and many others will give you any advice in the future if you keep doing that. Sorry mods but it pi..ses me of. Ban me if you think that I'm way out of line.
Edit; part of the purpose of a forum is to help the one with the question but many with similar problems read the posts as well. If the solution is not brought for a day it only helps one and that is waste of the many that looks for an answer. @Fatic you are banned from my list until you finish your first post.

 

[ppn]

you run the game of choice,, and while you play adjust +20 Mhz every hour until it crashes and then rolllback 20 Mhz. but have to unlock the TDP to the max, otherwise it will constantly throttle. my card drops to 1700 Mhz at the default 175 watt setting. to run 2000 the gpu needs 45% more power.

 

[Sasqui]

Still, Heaven has seen its day, and it is past.

As a quick artifact test, it's awesome The full 3Dmark suite is where you really find out about stability. I bought 3DMark on Steam for on the cheap and it's very much paid for itself testing both a 1070 Ti and Vega 64

 

[Solaris17]

As a quick artifact test, it's awesome The full 3Dmark suite is where you really find out about stability. I bought 3DMark on Steam for on the cheap and it's very much paid for itself testing both a 1070 Ti and Vega 64

That is true, I haven’t had it lock up or crash but it tells me by way of bright green squares all over lol.

 

[EarthDog]

I keep forgetting that I use the Hwbot version which just has Easy and Extreme modes (both of which are CPU tests these days). It can be configured to be a good/better GPU test, the normal version you guys typically use.

 

[Sasqui]

That is true, I haven’t had it lock up or crash but it tells me by way of bright green squares all over lol.

Oh I've seen my screen and PC do things that didn't seem possible, lol. The frustrating part is if you haven't saved an OC profile, you gotta get the system started again and work clocks and voltages back up to where you were.

And there always seemed to be some certain parts in timespy where one or the other card would crash, and crash hard.

 

[John Naylor]

It matters what Gigabyte card ... the reason that different models don't cost the same is some are built using better components than others. This usually results in better OC ability but Boost 3, is nerfing what could I think be higher OCs on the better componentry. On the 1st 8 TPU 1080 To reviews done here on TPU, the fastest delivered about 5% more fps manually OC'd than the slowest 216.r for the reference ... at that time, the fastest AIB card tested was the MSI Gaming X Trio at 226.6 and the slowest AIB was the EVGA Ultra X.

One thing I usually do is I look at what authors doing reviews on web sites gave done and 'start there:

https://www.guru3d.com/articles_pages/gigabyte_geforce_rtx_2080_gaming_oc_8g_review,29.html

Setting the following on the Gigabyte 1080:

• Temp Target 88 Degrees C
• Power Limiter: 108%
• CPU clock + 130 MHz
• Mem clock +950MHz (x2 DDR)
• Voltage +100%
• FAN RPM default

Results in:
Core Clock: 1645 MHz
Boost Clock: 2000~2100 MHz
Memory Clock: 15958 MHz

MSI Afterburner Image : https://www.guru3d.com/index.php?ct=articles&action=file&id=44044

And yes, they did a lot better than OC Scanner results:

Time Spy results :
Gigabyte 2080 Gaming OC scored 10600
Gigabyte 2080 Gaming OC scored 10837 w/ OC Scanner (+ 2.24% over stock)
Gigabyte 2080 Gaming OC scored 11136 w/ Manual OC (+ 5.10% over stock)

https://www.guru3d.com/index.php?ct=articles&action=file&id=44049

 

[Fatic]

Thanks for answers.Im using manuel OC on my card.
Clock: +120mHz
Memory: +350mHz
My card is working 70-72 degree celcius.

 

[LacklusterOG]

Some tips regarding oc scanner. If you want higher clocks, use msi afterburner beta 9. On my 2080ti, beta 9 hits 2225 on oc scanner, will game at 2190, them level off to 2175 once it heats up. Beta 12 will only get me up to 2120. You can upgrade versions after you get the better clocks and retain your profile. Another thing is to scan when your GPU is as cold as possible, run fans/cooling at 100%. My card will not clock nearly as high manually, only to about 2115. I also flashed my card bios for a higher tdp draw. This helps as well.