This was posted 4 months 15 days ago, and might be an out-dated deal.

Related
  • expired

PowerColor HellHound RX 7900 XTX 24GB GDDR6 Graphics Card $1249 Delivered ($0 VIC, NSW, SA C&C) @ Centre Com

1170
PowerColor HellHound RX 7900 XTX 24GB GDDR6 Graphics Card $1249 Delivered ($0 VIC, NSW, SA C&C) @ Centre Com
Go to Deal
Bruceflix on 18/05/2025 - 15:34 centrecom.com.au (1944 clicks)
Last edited 22/05/2025 - 19:41 by 1 other user
FAF150

Possibly ATL for a high end GPU. Hard not to recommend at this price.

Quick Specifications
General
Chipset Manufacturer
AMD
AMD GPU Generation
Radeon RX 7000 Series
Model
Radeon RX 7900 XTX
Interface Type
PCI Express 4.0
Graphics Card Memory
24GB
Memory Type
GDDR6
Memory Interface
384-Bit
Output
HDMI
DisplayPort
Digital Maximum Resolution
7680 x 4320
OpenGL Version Support
4.6
Recommended PSU (W)
800W
Power Connectors
8-pin x 2

Surcharges: 0% for bank deposit, Afterpay & Zip Money. 1.2% for VISA / MasterCard & PayPal. 2.2% for AmEx.

Free shipping excludes WA, NT & remote areas.

Computing AMD Computer Component End of Financial Year Sale PowerColor Video Card

Related Stores

Centre Com
Centre Com

closed Comments

  • spriggan43 on 18/05/2025 - 15:38

    This 7900xtx or 9070xt ?

    • Aarent on 18/05/2025 - 15:43
      +32

      imo 9070 xt

      • Bruceflix on 18/05/2025 - 15:45
        +4

        Really depends on what games you play and what resolution.

    • scud70 on 18/05/2025 - 15:50
      +19

      9070XT with way better upscaling ( they need more games to use it but it should be the way forward) and better ray tracing, uses bit less power. If you need the vram then 7900XTX. In Aus the 9070XT isn't as overpriced as other countries but it's still over what it's RRP is meant to be… unlikely to come down soon but 5070Ti's are starting to come down in price slowly so some pressure on AMD to keep it's "value for money" status.

      I have a 9070XT i got at RRP and it's a very good card and a great upgrade for me going for 6800XT.

      • Bruceflix on 18/05/2025 - 16:02
        +9

        9070XT with way better upscaling… and better ray tracing

        But if you cared about either upscaling or ray tracing you'd be better off getting Nvidia. And for pure raster the 7900xtx can hold its own, especially when overclocked.

      • penguinswillrise on 18/05/2025 - 17:08
        +1

        What's RRP? $1149 for the XT and $999 for the non-XT? Btw great to hear as I'm pondering an upgrade from 6800 non-XT.

        • tmr3 on 18/05/2025 - 19:02
          +8

          $1049 for the 9070, $1139 for the 9070 XT. 9070s are available at RRP, 9070 XTs are not ($1249 for the cheapest models I can see).

          I upgraded to a 9070 XT (PowerColor Red Devil for $1299) from a 3070.

          Out of the available cards in this performance tier 7900 XTX, 9070 XT, 5070 Ti), I'd be more likely to recommend a 5070 Ti for $1375 with eBay Plus, or $1409 + Delivery without.

          Unless you have a specific use case for the extra VRAM of the 7900 XTX, I'm of the opinion that the stronger feature sets of both the 9070 XT and 5070 Ti make them a better pick for most. The 5070 Ti is the best of the trio in that regard, and at current prices it's practically a no-brainer to pick it over a $1249/$1299 9070 XT.

          Head to head for the 7900 XTX vs 9070 XT only, I'd probably still pick a 9070 XT personally.

          • FujinShu on 19/05/2025 - 08:01

            @tmr3: Unless you want to use SteamOS.
            The RX 9070 XT is the best GPU for Steam machines.

          • WhatWhoWhyUsername on 20/05/2025 - 13:17

            @tmr3: How's the upgrade over 3070? That is my current GPU and deciding whether or not to upgrade also.

            • tmr3 on 20/05/2025 - 16:27
              +1

              @WhatWhoWhyUsername: For the games I play, it's typically been somewhere between about 1.7x and twice as fast. I've been using 1440p 165Hz monitors, but picked up a really cheap Samsung G6 OLED which is 1440 360Hz and while a 9070 XT is still not maxing out every game at that full refresh rate, it's definitely much more capable of utilsing it.

              Looking at my history of GPU upgrades, 70%+ has been roughly the performance level I've found acceptable, and I wanted to get it at least for similar cost to my 3070 after inflation. A 4070 Ti Super or 4080 (Super) last year probably would have been objectively better value but I wasn't in the market for a GPU at the time.

              I used to get really horrendous stuttering and other issues in the recent CoD games for example on my 3070 and that all went away completely once I swapped to the 9070 XT. I was able to practically max out the settings and still get significantly more performance. While the objective RT performance still isn't as well rounded as say NVIDIA's 50 series GPUs, there are some games where I've been more likely to turn RT on than with my 3070 simply because of the higher overall performance level. A 5070 Ti would still be faster there though of course.

              Not that I've done any A/B testing but FSR 4 feels at least as good as DLSS 3 ever was, subjectively I feel like it's perhaps slightly better at maintaining texture detail and limiting ghosting. To use CoD as an example again, there were plenty of times I'd notice ghosting or fuzziness in those games with DLSS (e.g. red dot sights leaving trails, or ghosting on particle trails/shadows) but those things are significantly less apparent with FSR 4. I didn't get a lot of time with DLSS 4 but there are many reputable comparisons there. I hope game support continues to improve but Optiscaler has done an okay job in the interim. One annoying thing is a game or driver update means you have to re-enable FSR 4 in the driver, at least for CoD.

              It's a higher power draw card that my 3070 (EVGA XC3 Ultra which I kept undervolted/clocked to around 1800 MHz), but the extra size and bulk of the Red Devil has meant I've not really had to play around with clocks or settings at all to keep it under control.

              Overall I've been quite pleased with it. While I do think a 5070 Ti at current sale prices makes more objective sense for most, I wanted to switch back to an AMD GPU provided a competitive option in this price tier existed, so the 9070 XT was an easy pick for me over the 5070 Ti. Part of it is being a bit fed up with NVIDIA's business practices, part of it is the annoyances I had with my 3070, and part of it is that I'm thinking of switching to Linux as my daily driver.

              If you're not having any major issues with your 3070, my default recommendation is always wait - whether it's for prices to continue to fall or for new products. I elected to upgrade as a final step in essentially maxxing out my current platform (5800X3D on AM4), so I'm likely holding onto this build for a few years and starting from scratch when the time comes.

              Hope that helps!

              • WhatWhoWhyUsername on 20/05/2025 - 19:04

                @tmr3: Appreciate the reply, I'm fed up with NVIDIA too with constant driver issues and am experiencing random black screen. I don't care about RT, I think the AMD is better value for me right now. Not quite in a rush and willing to wait a little longer hoping the prices may drop back closer to RRP.

                • tmr3 on 20/05/2025 - 19:31

                  @WhatWhoWhyUsername: Yep, waiting is the right call I think. Hopefully there are some good EOFY deals coming up!

        • scud70 on 19/05/2025 - 07:28

          Yep that's right, I got the Reaper model for $1149 a week or two after release. It'll be a very good upgrade from 6800, but I'd def recommend getting one that's close to the RRP. Unfortunately for AMD as tmr says there are 5070tis in the same ball park, when that happens with DLSS 4 id recommend that, I personally don't buy nvidia but that's just a personal preference, I'd recommend them to others if it's better value (although this is marginal and if you get a 9070XT for low 1200s I'd save the money vs nvidia and get that).

    • jasswolf on 18/05/2025 - 16:11
      +9

      The 5070 Ti for under $1400.

      • Bruceflix on 18/05/2025 - 16:36
        +1

        With Doom Premium

    • sludo on 18/05/2025 - 18:48
      +2

      If you’re playing 4K this if you’re playing RT games and 1440p the 9070 xt

    • mckayver on 18/05/2025 - 20:29

      I have 9070 xt and very happy with it.

    • Franc-T on 19/05/2025 - 06:12
      +1

      I have this exact card, and whilst it is brilliant, id get the 9070XT due to better RT - if the same price.

  • Hornpub on 18/05/2025 - 15:47
    +8

    Delicious 24gb

  • Jaspa7 on 18/05/2025 - 15:51
    +4

    recommended psu 800W - RIP :'(

    • stephendt on 18/05/2025 - 16:17

      They can run on a good 500w PSU with some tuning

      • Bruceflix on 18/05/2025 - 16:27
        +3

        I had the 7900xtx Saphire Nitro+. That used to pull 400w out of the box and 460w overclocked.

      • OzzyBrak on 18/05/2025 - 17:39

        How do you know if you are under powered?

        • WhyAmICommenting on 18/05/2025 - 18:05

          If you hear loud buzzing in the PSU then you are not underpowered ;)

    • Dvbargain on 18/05/2025 - 17:25
      +2

      I have a 7900xtx/5800x3d rig running for years on a Corsair RM750

      • OzzyBrak on 18/05/2025 - 19:00

        I'm still on my old Corsair HX750i from 2017, it's the oldest component in the box.

      • TheBargainDog on 19/05/2025 - 00:00

        Very similar setup here with a Nitro+, absolutely no issues for over a year.

    • netsev on 24/05/2025 - 16:51

      You don't have to use the recommended, you can get by with a good 650 psu I reckon.

      Just roughly adding up, this gpu 350w + 150-250w (cpu/mb/mem/hdd/ssd/etc) comes to about 500-600w under load.
      Generally you don't want your psu to be running close to 100% you need to give it some headroom to be efficient.

      If you game regularly you want at least a good 650w psu, preferably 700+.

  • nobro25 on 18/05/2025 - 15:56

    Is this good enough for GTA VI?

    • Aureus on 18/05/2025 - 16:00
      +12

      Game isn't coming out on PC for years, so bit early to be looking for a GPU. But I am sure it will be fine for GTA6.

      • Bruceflix on 18/05/2025 - 16:03
        -4

        But I am sure it will be fine for GTA6

        On low at 720p..

        • Xukeepa on 18/05/2025 - 16:08
          +3

          if the current gen console can run it at decent quality, 7900xtx can do it too.

          • jasswolf on 18/05/2025 - 16:12

            @Xukeepa: So far it's 1440p30 using some aspects of FSR3.

      • siggy213 on 18/05/2025 - 16:14
        -1

        May 2026 assuming they don't push it back again

        • Shaun Prawn on 18/05/2025 - 16:22
          +9

          I think that is only for consoles… And PC release will probably come some time later…

        • Chaubie on 18/05/2025 - 17:55
          +1

          yeah its going to probably be 2027…

    • netsev on 18/05/2025 - 20:48
      +1

      This card is equivalent to a 4080 it'll run everything fine. If AMD enable FSR4 on these…then it'll be a real bargain and better than 9070XT

    • Xukeepa on 18/05/2025 - 17:37
      +3

      Near Zero would not be worth the effort unless you already have distribution in China.

    • Franc-T on 19/05/2025 - 06:13

      Not cuda compatible so none

      • Forth on 22/05/2025 - 15:20

        Zluda works fine on AMD cards :)

  • superroach on 18/05/2025 - 16:45
    +1

    Good price for 24gb!

  • Wally Simmonds on 18/05/2025 - 17:02
    +1

    Surprised by the lack of 9070xt deals.

    • FujinShu on 18/05/2025 - 17:17
      +5

      High demand, that’s all.
      Surprisingly, compared to previous generations, it’s now the inverse, where Nvidia’s getting all the deals and AMD has none.
      Although if SteamOS and Gamescope get stable support for Nvidia GPUs, then I’ll probably just buy an RTX 5070 Ti for SteamOS.

    • Logical on 18/05/2025 - 22:19

      picked up the sapphire pure white OC 9070XT for $1288 just the other day

      • bAps on 19/05/2025 - 11:13

        i.e. paid higher than RRP

        • Logical on 20/05/2025 - 00:25

          business expense so around 40% back at the end of the day

  • vid_ghost on 18/05/2025 - 17:20
    +1

    24GB NICE

  • Dack Smith on 18/05/2025 - 17:52
    -4

    Great deal

    This pretty pisses all over the NV 5000 offerings which are just popular but expensive paperweight explosives.

    But I suppose rabid nv fanboys beg to differ

  • lordezekiel on 18/05/2025 - 18:14
    +4

    ~$54/GB

  • Relsierk on 18/05/2025 - 18:58
    +5

    If you plan to run AI applications, this 24 GB will be necessary. My current 16 GB on the 7900 GRE is now borderline on the minimum limit.

    • su27km on 18/05/2025 - 19:07
      +1

      Why not purchase Nvidia 5060ti 16G

      • Wally Simmonds on 18/05/2025 - 19:45

        Yeah you can run llms and stable diffusion on 12gb nvidia cards

        4060ti/5060ti 16gbs are decent for this

      • Forth on 22/05/2025 - 15:25

        5060ti? Your better off with a second hand RTX 3090 running Ai and LLMs for $1k.
        7900XTXs using LLMs are pretty fast. Now with better ROCm coming.

    • FujinShu on 18/05/2025 - 21:53
      +8

      If you plan to run AI applications, just get literally any Nvidia GPU with 24GB memory (doesn’t have to be GeForce).
      ROCm is not that great, and RDNA is very much a gaming-focused card with not-great non-gaming performance anyway.
      Good for gaming, terrible for anything else.

      • whatisk on 19/05/2025 - 06:48
        +1

        Really wish there was a 24GB Nvidia at this price for AI…

        • FujinShu on 19/05/2025 - 07:56
          +1

          If you’re willing to buy older GPUs, you can get enterprise GPUs with at least 24GB VRAM.

          • whatisk on 19/05/2025 - 09:56

            @FujinShu: Age doesn't particularly bother me. Just getting tired of generation times with my RTX3060 12GB.
            Are there ones that would be recommended?

            • FujinShu on 19/05/2025 - 10:58

              @whatisk: There’s the Quadro P6000, which is a Pascal GPU with 24GB VRAM, which I found a listing for $900 (although you can probably find it cheaper in other listings).
              Even though it’s Pascal, it’s still decently performant especially for CUDA-based AI, and it has 24GB VRAM.
              If you’re willing to sacrifice 2GB memory, you can get a modded RTX 2080 Ti, which is a Turing GPU with tensor cores, with 22GB memory. That’s around $1000, but you can also just buy the 11GB base graphics card itself and either mod the memory yourself or get someone else to do it.
              Of course there are other options that are cheaper or better or both, depending on your needs, and this is just what I’ve found from an eBay search.

            • idonotknowwhy on 19/05/2025 - 12:27
              +2

              @whatisk: I wouldn't recommend this. Sounds like what everyone was doing 2 years ago.

              Those older cards like the P40 lack FP16/BF16. You'll end up with less options, and slower speeds than an Intel/AMD GPU.

              12GB

              If you haven't already, try Qwen3-30B-A3B with all layers offloaded to the 3060, then —override-tensor to move most of the experts to your CPU.

              Claude/ChatGPT can help you tweak the regex to put some of them back on the GPU for more performance. You can see people discussing it here.

              • whatisk on 19/05/2025 - 13:33

                @idonotknowwhy: Thanks for the info.

              • FujinShu on 19/05/2025 - 16:25

                @idonotknowwhy: Do slower speeds matter if the faster cards aren’t optimised anyway because of CUDA usage?
                But yes, if your program utilises open standards of ROCm, then you’ll be fine.

                • idonotknowwhy on 20/05/2025 - 19:48
                  +1

                  @FujinShu: Sorry just to clarify, I'm certainly not advocating for the mess that is ROCm (I got lured in by the lost cost to rent MI300X instances, discovered why they're so cheap lol. Most AMD GPU users who only want to run inference end up using Vulkan.

                  And you're right, ROCm (AMD), Sycl (Intel/Nvidia) and Vulkan(anything) are all a lot slower than cuda.

                  The only thing that comes close is OpenVino, but it's got limitations too (Intel GPUs only reach ~RTX4060 performance, BF16 doesn't work yet, tensor-split doesn't work, etc)

                  I'm excited for the new 24GB and 48GB cards Intel just announced.

        • xrailgun on 19/05/2025 - 23:27
          +1

          modded 2080 TIs 22gb are around $600 AUD on taobao, or about $900 on aliexpress/ebay.

          regular used 3090 (24gb) are around $800-1000 AUD on marketplace

          • FujinShu on 20/05/2025 - 13:51

            @xrailgun: I don’t trust FB Marketplace here.
            Unless they offer PayPal, which at least has recourse if your order goes wrong.

          • whatisk on 21/05/2025 - 06:46

            @xrailgun: Cheers. Will do a bit more investigation. I do image and video generation rather than LLM usage.

    • Franc-T on 19/05/2025 - 06:15
      +5

      Bad take Im afraid. If you want to use AI, get a nvidia card. Radeons are such a pita to try to use and you will constantly have issues.

      I have a 7900XTX and bought a 3090 for AI stuff as Radeons. I dont even try to run AI stuff on the 7900XTX

      • OzzyBrak on 19/05/2025 - 09:55
        +1

        CUDA is king

        • FujinShu on 19/05/2025 - 10:58
          +1

          Hopefully CUDA can be opened up either by China or the EU.

          • OzzyBrak on 19/05/2025 - 16:55

            @FujinShu: Hardly seems fair. I’m all for open source, but forcing proprietary software to be open strips away a lot of the value a company has invested in to build.

            A better approach would be an open standard, allowing ML packages to be device agnostic, something like pytorch_any that runs the same workload on any GPU. Let’s be honest though, everything would still run faster and better on Nvidia hardware anyway.

            • FujinShu on 20/05/2025 - 13:36
              +1

              @OzzyBrak:

              Hardly seems fair.

              Fairness went out the window as soon as corporations started abusing their stakeholders and compromising governments for the sake of their shareholders.

              At least by bringing in CUDA as an open standard, it will allow for further advancement and competition from all AI accelerator designers by allowing them to run the vast backlog of Nvidia-only programs, thus getting everyone on (mostly) equal footing.

              And perhaps the best part will be Nvidia actually being forced to actually provide value beyond just being the only option, lowering prices and making computing more accessible.

              Controlled competition fosters collaboration and advancement, but uncontrolled competition results in the oligarchic crapshow that is America.

              • OzzyBrak on 20/05/2025 - 18:34

                @FujinShu: The opposite is actually true. America is responsible for most tech innovation.

                Forcing companies to lose their competitive edge by removing their ability to protect their intellectual property reduces innovation and investment.

                Having frameworks be agnostic is one thing, forcing the secret sauce out of CUDA reduces the incentive to invest money into making it the best.

              • idonotknowwhy on 20/05/2025 - 19:53

                @FujinShu: Gotta agreed with OzzyBrak on this one. I'm all for opensource, and really want Vulkan to catch up for inference and OneAPI/OpenVino to be better documented and even port projects over to XPUs, but Nvidia took a risk investing so much in CUDA. We need to alternatives to catch up, not to force Nvidia to give it away.

          • Franc-T on 19/05/2025 - 16:56
            +1

            @FujinShu: There are altenatives, but CUDA is pretty entrenched

      • Forth on 22/05/2025 - 15:29
        +1

        I did the same, great minds think a like.
        3090 Ai, LLMs
        7090XTX for gaming, LLMs

        • Franc-T on 22/05/2025 - 18:36

          Nice!

          • Franc-T on 26/05/2025 - 17:14

            @Franc-T: 3090 keeps overheating, stupid Acer dont have anywhere how thick thermal pads were and its a bugger to work out.

            Bought a 5060 (16GB) to run in meantime……..it runs so cool its not funny

  • thok2024 on 18/05/2025 - 23:07
    +2

    Solid deal. Been watching this site since my buddy recommended it almost a year ago. This seems to be the lowest price ever. Just ordered it using the code for my first PC build in years! :)

  • Bruceflix on 20/05/2025 - 13:33

    FAF100 has been extended to May 31

  • irev1 on 20/05/2025 - 17:55

    I have a 7900XTX and love the card however you can get the 9070XT for the same price $1300 here https://www.centrecom.com.au/sapphire-pulse-rx9070-xt-16gb-o…

    The 9070XT is still the better card/package even tho slightly less vram and raster performance. Day to Day and over the next 3-5 years the 9070XT is the better choice

  • beniSpurdo on 21/05/2025 - 08:00

    Enticed by this if it didn't have the most horsedung tasteless gamer aesthetic cooling shroud

    • kickasspeanut on 21/05/2025 - 14:39

      You're supposed to be looking at your monitor, not your GPU.

  • Tanvry on 22/05/2025 - 19:14
    +1

    haah it's $150 off now $1249

    • Bruceflix on 22/05/2025 - 19:42

      Updated. Thanks

      • Tanvry on 26/05/2025 - 14:46

        it's now $1349 cuz they raised the price on the GPU by $100

Login or Join to leave a comment
Login Join