I have the exact same model vega 56, I flashed the vega 64 bios onto it and it worked a charm. Only downside is that it started to sound more like a jet engine.... very very loud lol
One fun factoid is that the Vega chips are quite tunable even if they don't overclock all that great. It's possible to reduce the power limits on the card without severely affecting the performance and get substantially better temperatures. Did the same on a few other AMD cards from the R9 200 generation and an RX 580. GCN was a very interesting architecture.
I had a sapphire blower v64 with Samsung hbm, it could easily do 1100 on the memory and with a block I had purchased off of aliexpress it could reach up to 1700mhz or a bit higher on the core Almost 20% performance increase over stock but power figures off the charts It also used to undervolt pretty well, with maintaining stock frequency and undervolting I could almost cut the power draw to around 60% stock without losing a single % performance Really cool card, never should have sold it 😢
@@ProYamYamPC yeah, though vega was never intended to be used as a gaming gpu. Amd had made vega specifically compute workloads and for apple so they could use it in their macintosh’s, mostly being good only in compute. They just run games by brute force. The entire gcn lineup was good for compute but really bad for graphics workloads. Since Amd made too many cards, for the gaming side of things, amd expected the vega architecture to clock much higher than what they actually clock at, but the cards that came out of the fab were simply not good enough, but they had no choice but to release what they had.
Run mine in my misses pc, does the job great still. least for all she plays. Prob said before, but mine has a 64 bios and a morpheus 2 cooler. Such a fun card.
Wanna know something odd. I built my wife a computer and put a vega frontier edition gpu in it. That card runs games faster than my radeon vii and my arc a770. She has a ryzen 5700x3d in her computer as well.
@theanglerfish that's what I thought. On MW3 it runs at the game at 1440p with fsr max setting at 153 fps. The radeon vii runs the game with the same settings at 110fps and the a770 is around 100fps same settings.
@@bryanwages3518 A770 don't have so much memory bandwith which really helps in this game but don't understand that difference between these radeons but frontier have about 200 cores more also tmus which these are texturing units but that difference is so much wide this did not makes me any sense
@theanglerfish I absolutely agree. It makes absolutely no sense. I will say I do have AMD SAM support enabled for both the frontier and radeon Vii card. You can enable it with a registry change. Rebar is enabled for the arc card aswell.
I run my Vega 56 daily underclocked to 1138 Mhz core / 900 Mhz mem / -100mV and it runs very cool, quiet, and very power efficient pulling 120W on benchmark app Also you should try HBCC, they will become this gpu saving grace in a vram limited scenarios, in game or others like AI workloads... I made one video comparison case of that HBCC
I think more of a 5700XT for budget 1080p gaming..But this shows how far from E-waste old tech actually is. And its kinda scary to realize how much FOMO or fear in general dictates our decisionmaking.
I got mine for 75$, i just upgraded yesterday, too bad I have an i7 2600, so I essentially had the same fps with my rx 470, but on 2560x1600 because if I go lower, the card doesn't get utilized at all lol.
If u had a card with samsung hbm then u would have a winner on your hand. my v56 and v64 with samsung hmb could go at 1100 to 1200. Look for those, asus and sapphire cards.
Why so using old graphic card but modern like efficient power and higher clock and more cores than it you can get RX 7600/7700 XT left is cheaper than 7700 XT. I had a RTX 3060 was great for 1080p but I use a RX 7900 XT only for VR however I can play any games with 1440p everything is future to make a graphic card that can handle 4K but it won't be enough such as 540hz or other its bandwidth limited. Like I have a new monitor is 1440p 180hz.
Oh, i have a question based off today's hype. Unsure if nvidia stated this, but someone was banging on about how nvidia didn't need as much ram as amd as their ram is faster, so it means it needs less lol. I don't really believe that. If that was the case then surely hmb2 would of been doing the same thing, but that was never said back in vega days. I don't call that ever being pushed.
It can be done with HBM. even the original Fury X can do it. But in case of Fury X AMD need to tune it per game basis. And AMD historically did not like to do per application optimization. So they abandon Fury X and create HBCC for Vega. That way AMD no longer need to do per application optimization but the issue with HBCC is VRAM utilization must always exceed the amount of VRAM available on the card. For example vega have 8GB so you must ensure VRAM usage to stay above 8GB at all time. Back then even 4k can't really fill 8GB all the time in many games. If VRAM usage drop below 8GB there will be performance issue. So AMD recommemds HBCC to be turned off by default in driver settings.
@@ProYamYamPC isn't that hynix HBM supposed to be better than samsung HBM? i remember even on professional card the Hynix HBM usually will have much higher clock than samsung HBM.
Vega was bleeding edge. but what we saw with nvidia vs AMD is bleeding edge does not guarantee success. HBM might be more modern and more impressive than old GDDR memory but in the end it end up preventing AMD to do better. HBM was so expensive that AMD have to limit Vega 64/56 to 2 stack only which in turn limiting the bandwidth available to the card. with Radeon 7 AMD finally using 4 stack of HBM and provide enough bandwidth for the card. AMD keep the amount of CU the same as Vega 64 and yet with improved bandwidth the Radeon 7 finally on par with GTX1080Ti performance.
@@gejamugamlatsoomanam77165090 doesn't exist and 4090 is used by 1% of Steam users (August 2024 survey). The vast majority of people couldn't care less about halo cards.
