RX 7900 XTX Hashrates and Overclocking

How well is the top end AMD GPU mine Cryptocurrency well obviously it depends On the cryptocurrency you were talking [Music] About today we are taking a look at the Hash rates for the RX 7900 XTX this particular model is the ASRock Phantom gaming overclock Edition it Features 96 compute units which can Reach a boost clock of 2,615 mehz out of the box which in Theory should benefit core intensive Algorithms on the memory side of things It features 24 GB of gddr6 clocked at 2500 MHz on a 384bit bus which results In a bandwidth of 960 GB a second this Can be increased to over 1 terabyte a Second with overclocking powering the GPU are 38 pin pcie power adapters which Can deliver well over 400 watts to the The GPU with a recommended power supply Of 1,000 wats for testing I utilized the Corsair hx200 watt PSU and at these Loads it provides 87% efficiency so you Can keep this in mind as all power is Calculated at the wall to be clear the Riser and the card are powered by a Separate PSU from the rest of the system For accurate readings cooling the card Are three fans over a nickel plated Copper base while they advertise premium Thermal pads for the memory modules they Regularly go over 100° C when pushed to 2600 MHz for testing operating systems

I show You how To Make Huge Profits In A Short Time With Cryptos! I show You how To Make Huge Profits In A Short Time With Cryptos! Welcome to the Future of Money

Hios stable and beta images were tested However the overclocks would not apply Switching to MMOs V4 resolved this issue Out of the box and allowed full control Over core clock voltage and memory clock Now that we have the overview out of the Way let's discuss stock hash rates and The best minor out of the box for these Units before choosing which minor to use For the overclocks on auto Leos V2 which Is a memory intensive algorithm the RX 7900 XTX worked on four miners bz minor Team red minor nanom minor and SRB minor With team red Miner outperforming all The others at 193 mahash a second at 416 Watts resulting in 46 mahash a second Per watt on beam has which utilizes a Bit of memory and core only LOL minor Would work and resulted in 47.3 Solutions a second at 410 Watts Resulting in11 Solutions a second per Watt on Blake 3 alium which is a core Intensive algorithm three miners worked Bz minor LOL minor and SRB minor with SRB minor taking the lead at 3,43 Megahash a second at 419 Watts resulting In 8.12 Mah a second per watt on Blake 3 Ironfish which is another core intensive Algorithm four miners were supported bz Minor SRB minor LOL minor and team red Minor with team red Miner taking the Lead at 3.42 ghash a second at 420 Watts Resulting in 0.072 gigahash a second per watt on DX

Solve which uses a bit of the core and a Bit of the memory only one minor worked Bz minor which got 7,660 Hash a second at 395 Watts Resulting in 19.4 has a second per watt On Etc hash which is a memory intensive Algorithm 7 minor work bz minor LOL Minor SRB minor team red minor G minor Phoenix minor and Nano minor with nanom Miner taking a large lead over the pack With 110 mahash a second at 414 Watts Resulting in26 mahash a second per watt This was verified on the pool however it Is important to mention that it does Fluctuate and hash rate more than the Other miners but even at the low results It would hit 105 mahash a second and Near the top end 115 mahash a second Hence the average of around 110 mahash a Second on eash which is another memory Intensive algorithm five miners worked Lol minor team red minor SRB minor G Minor and nanom Miner with nanom Miner Once again taking the lead with 110 MHz A second at 46 16 Watts resulting In26 mahash a second per watt on eb3 Which is a memory intensive algorithm Two miners worked lol minor and SRB Minor with SRB minor taking the lead With 102.9 mahash of a second at 414 Watts resulting in24 mahash a second per Watt on Carlson hash which is a core Intensive algorithm three miners were Supported bz minor SRB minor and LOL

Minor with LOL L Miner taking the lead Getting 2,319 mahash a second at 419 Watts Resulting in 5.53 megahash a second per Watt on capow which uses a bit of memory And core four miners were supported wild Rig nanom Miner bz minor and SRB minor With SRB minor taking the lead mining at 57.1 2 mahash a second at 411 Watts Resulting in13 mahash a second per watt On M hash which uses a bit of core and Memory only SRB minor was supported with 117 mahash a second at 410 Watts Resulting in 28 mahash a second per watt On Nexa proof of work which uses a Little memory and core only LOL minor Was supported with 118.1 3 mahz a second At 48 watts resulting in 28 mahash a Second per watt on Pyon hash which is a Core intensive algorithm two miners were Supported SRB minor and LOL minor with LOL minor taking a hefty lead getting 7,880 mahash a second at 416 Watts Resulting in 18.94% mahash a second at 420 Watts Resulting in 14.7 mahash a second per Watt on Shaw 3D which is a core Intensive algorithm two miners were Supported bz minor and SRB minor with SRB minor taking the lead at 139 mahash A second at 419 Watts resulting in 3.12 Mahash a second per watt on Shaw 512 256d which is a core intensive algorithm Three miners were supported bz minor SRB

Minor and LOL minor with LOL Miner Taking the lead at 1977 mahash a second At 419 Watts resulting in 4.72 Mah a Second per watt finally on zel has which Uses memory and core two miners were Supported G minor and LOL minor with LOL Miner taking the lead at 103.5 Solutions A second out 414 Watts resulting in.3 Solutions a second per watt before I get Into overclocking I did test dual mining To determine which algor alms have the Highest impact on hash rate per watt First testing core intensive algorithm Impact on ET Blake 3 alium had the least Impact with 23 mahash of second per watt With Carlson hash impacting Etc hash the Most taking it down to. 21 mahash a Second per watt second I also tested Memory intensive algorithm impact on Hash rates on Blake 3 alium where you Can see ET has the least amount of Impact at 4.47 Mah a second per watt With Etc hash showing the most impact Taking alium down to 3.82 mahash Second Per watt dual mining in and of itself Has many factors in overclocking options Due to the time commitment needed I did Not address any overclocking while dual Mining you can get a pretty good Starting point from the overclocking I Will now cover for single algorithms it Is important to note that all my power Is recorded at the wall for sanity Checks I did reference hash rate. o in

Comparison to software readings as well That being said in most cases I have Found that you can get better Performance than reported on hash rate. Or in some cases there was no way to get Close to the efficiency as the wattage Reports in software are incredibly off From what is showing at the wall and the Utilization of different miners report Differently another note is that there Are variances between different makes And models Therefore your mileage may Vary but I will cover the basic Guidelines before providing specifics as We know from the specs this RX 7900 XTX From ASRock has a max boost clock of 2,615 MHz and the memory is clocked at 2500 MHz out of the box it is safe to Assume that the memory overclocks will Be consistent across all makes and Models outside of potential cooling Issues in the case of the gddr6 on board The max overclock is 2600 MHz which Translates to 1300 MHz in Linux when Running memory intensive algorithms this Should always be set to 1300 MHz or 2600 MHz in Windows a nice feature of the 7000 series is its ability to Automatically clock the memory down to 97 MHz when the memory is not being used This means that things like spart memory Tune on MMOs are not necessary in fact When using compute heavy smart memory Set to low it will clock the memory to

772 MHz resulting in higher power Consumption over not touching the memory At all finally on algorithms that Utilize core and memory there was some Variants where a lower clock of memory Provided the best efficiency these also Tend to benefit from Smart memory tune In MMOs a good starting point for Algorithms such as kapow is 1050 MHz in Linux which translates to to 2100 m in Windows with all of that out of the way Let's discuss the results for the memory Intensive algorithm Aldo Leos V2 which Most famously mins ergo it was very easy To find the best overclocks studying the Core voltage to 680 MHz core clock to 1225 MHz and memory clock to 1300 MHz we Dro the power draw to 286 Watts at the Wall and increase the hash rate to 25.9 Mahash a second resulting in 71 mahash a Second per watt which is a 55% increase In efficiency over stock now just to Clarify once again here this sounds a Lot worse than it actually is up to this Point most of the numbers you hear Online are people looking at software These settings will result in sub2 200 Watt software readings which results in Over 1 Mah a second per watt but Obviously I wouldn't report that as fact Due to what is actually being used at The wall another note here when it comes To power is that you could calculate

Efficiency of the hx1200 being 87% at This load and remove the 133% from the 286 Watts resulting in 248 watts of real World usage and that results in 83 Mahash a second per watt finally on Additional notes Here is that you can Get better efficiency off 200 to 240 Volt circuits and Platinum rated power Supplies at over 50% utilization Continuing forward in these videos I Will not address these caveats for Brevity but I will note in the Description moving on to beam overclocks Which utilizes a bit of core and memory Means that it was important to test the Variance of efficiencies on both M and Core efficiency did not increase with Dropping the memory meaning beam does More heavily lean towards total memory Bandwidth therefore leaving the memory At 1300 MHz was per referred this Translated to the core as well where Dropping the core clock and voltage Resulted in poorer performance than Peak Efficiency core clocks of 2300 MHz with Voltage of 760 molt this resulted in 52.9 Solutions a second at 282 Watts for An efficiency of1 187 Solutions a second Per watt netting a massive 62% increase In efficiency over stock settings Blake 3 alium is the algorithm I utilize to Set the core intensive starting points And therefore have the most results for It it was found that a core voltage is

740 molts with a core clock of 2200 MHz And a memory clock of 97 MHz results in The best efficiency with 2,870 mahash a second at 212 Watts for An efficiency of 13.53 mahash a second Per watt netting a 66% increase in Efficiency for Blake 3 fish I started at The efficiency overclock for Blake 3 Alium I reduced the core voltage and Clock initially and discovered that it Increased efficiency this continued all The way down to 1900 mahz on the core at 720 molts on the clock leaving the Memory at 97 MHz anything lower than 1900 MHz resulted in a hang this gave me 21.2 ghash a second at 164 Watt for an Efficiency of .129 mahash a second per Watt which is a 78% increase in Efficiency Overstock clocks moving on to DX solve which utilizes memory and core It was important to test both memory and Core however it was quickly determined That dropping the memory overclock Resulted in a reduction of efficiency And therefore was left at 1300 MHz for The core clock dropping it down to 2,000 MHz and a core voltage of 730 molts Resulted in the efficiency uh at 7,850 h a second at 273 Watts resulting In 28.75 has a second per watt netting a 48% increase in efficiency on ET hash it Was pretty easy to determine the best Overclock setting which was setting the

Memory to 1300 MHz and dropping the core Clock to 1225 MHz at 680 molts this Resulted in 115 mahash a second at 288 Watts resulting in. 39 mahash a second Per watt netting a 49% increase in Efficiency this was a similar story on Eash as well utilizing the same Overclocks as Etc hash resulting in 115.2 mahash a second at 280 WTS Resulting in 41 mahash a second per watt Netting a 56% increase in efficiency Overstock clocks as you all know even ET Hb3 is similar to both Etc and ET hash However nanom Miner is not supported so The max hash rates are slightly lower at 107.8 meah a second at 2855 Watts resulting in37 meah a second Per watt netting a 52% increase in Efficiency over stock another core Intensive algorithm Carlson hash netted Similar overclocking results to Blake 3 Alium with the core voltage at 720 molts Core clock at 1900 megaherz and the Memory clock at 97 MHz the 7900 XTX got 1673 mahash a second at 164 Watts Resulting in 10.2 mahash a second per Watt for an increase in efficiency of 84% over stock kapow is a beast and Requires basically the high core and Memory clocks available within reason Going above 2300 MHz however did result In h so the final settings were a core Voltage of 740 molts at 2300 MHz on the Core and 1300 MHz on the memory for a

Hash rate of 59.7 Mah a second at 414 Watts resulting in .42 mahash a second Per watt which is only a meager 2% Increase in efficiency I messed with This one quite a lot and found little Success in improving the situation to The point where I would either lower the Clocks to reduce the Heat or avoid the Algorithm all together on the 7900 XTX Mem has is unique in that it does Utilize both memory and core while Favoring core this meant lowering the Memory clock to 1050 MHz resulted in the Best efficiency while cranking the core To 2400 MHz and the core voltage to 750 Molts gave me 9 mahash a second at 286 Watts resulting in 38 mahash a second Per watt for a 33% increase in Efficiency NEX a proof of work benefited From conservative clocks on both memory And core with a core voltage of 740 Molts core clock of 2200 MHz and a Memory clock of 1050 MHz for 10639 Mah a second at 301 Watts Resulting in35 Mah a second per per watt For a 22% increase in efficiency over Stock pyron hash threw another surprise Out as dropping the core clock resulted In better efficiency but going to 1900 MHz showed diminishing returns the best Efficiency settled in at 730 molts at 2,000 MHz on the core with a memory at 97 MHz for 5,650 Mah a second at 165 Watts

Resulting in 34.3 MH second per watt for A pretty significant 80% increase in Efficiency over stock Shaw 256 DT was Fairly simple as it mimics a few other Core intensive algorithms with a core Voltage of 720 molts at 1900 mahz with a 97 mahz underclock on the GPU clock for 4,280 mahash a second at 168 Watts Resulting in 25.4 mahash a second per Watt for a 73% increase and efficiency Over stock Shaw 3D utilized the same Overclocks as Shaw 256 DT for 94 mahash A second at 164 Watts resulting in 5.5 Mahash a second per watt for a 76% Increase in efficiency over stock Shaw 512 256d also used the same overclocks As the previous two getting 1380 mahat a Second at 173 Watts resulting in 7.9 Mahat a second per Wat for a 68% Increase in efficiency finally zel has Utilizes memory and core clock but Through a kink in the normal overclocks Even further where Peak efficiency on The M clock came in at 1200 MHz while Going lower had diminishing returns but Simultaneously ate up the core maxing Out the 2400 MHz at 760 molts giving 99 Solutions a second at 308 Watts Resulting in 32 Solutions a second per Watt for a s a smaller increase of 28% Increase in efficiency When comparing Results from hash rate. I achieved Better software results on both pyron Hash and Novo most notably however on

Radiant it was not possible to match the Efficiency results listed taking a look At overclocking performance for the RX 7900 XTX there are clear benefits in Core intensive algorithms thanks to the Memory clocking itself down when not in Use however this is not enough to make Up for the amount of power AMD pushed to Gain these overall results there was a Clear focus on gaming for this GPU and It has lost all the Allure of the Efficiency we saw with the RX 6000 Series perhaps this changes with the Rest of the 7,000 series which I will be Testing later so make sure you Subscribed to see that as far as memory Intensive algorithms are concerned due To the amount of additional power Required it is the same Story the RX 7900 XTX belongs in a gaming PC not in a Mining rig funny how this was achieved With engineering and not software blocks Like light hash rate from Nvidia I am Disappointed as you know I love big Buses but when the bus comes with a Bunch of baggage sometimes it's best to Just let it go compounding The increased Power problem is a lack of support the 7000 series has been largely ignored by Mining software developers and that is Sad in reality the high clocks should Result in better core intensive Algorithm hash rates than we are seeing Alas this is the same old story with AMD

As always people pay more for NVIDIA Because it has more support and Developers focus on Nvidia because they Have a larger install base maybe one day This will not be the case but for now it Is I cannot recommend the RX 7900 XTX at This time for mining I do enjoy the card Immensely though and it holds a special Place in my heart for the underdog Attempt as always additional review Materials and results will be available On son ofate tech. locals.com if you Would like to see a full tear down of This card with thermal paste and pad Replacement please let me know in the Comments I will see you next Tuesday

You May Also Like