• Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.
  • The forums have been upgraded with support for dark mode. By default it will follow the setting on your system/browser. You may override it by scrolling to the end of the page and clicking the gears icon.

RX9000 series choice

elano68

New Member
Joined
May 23, 2025
Messages
13 (0.57/day)
Hello again. I bought Nitro 9070xt for the gaming oc's price but Taichi is still in my mind :D

Asrock Taichi models are at the top in 3dmark steelnomad tests. Is this a coincidence or do the silicon qualities of this model always seem good? Should I stay with Nitro or should I add $90 more and buy Taichi?

I am very undecided. Please, I am waiting for your comments.
 
Last edited:
Hi mates. Ibought Nitro 9070xt for the gaming oc's price but Taichi is still in my mind :D

Asrock Taichi models are at the top in 3dmark steelnomad tests. Is this a coincidence or do the silicon qualities of this model always seem good? Should I stay with Nitro or should I add $90 more and buy Taichi?

I am very undecided. Please, I am waiting for your comments.

Unless your 9070 XT is a total dud, personally, I wouldn't bother. The Nitro is one of the best designs out there. It's entirely possible that the Taichi you get is a worse sample. Or it could be better, it's literally playing the roulette. No practical difference there would be worth $90.
 
Thank you for the information you provided.

What do you think could be the reason why taichi models are at the top?
 
Both the Taichi and Nitros are top cards, you got 1, stay with it unless if you are having a bonafied hardware issue like it being DOA.
 
Keep that Nitro for now, Get UDNA Taichi in 2026
It's really a 4 faced die roll when it comes to the Nitro, Taichi, Red Devil and the Top End XFX gpus
 
It's really a 4 faced die roll when it comes to the Nitro, Taichi, Red Devil and the Top End XFX gpus
Sure

If the OP is fascinated by the looks of the Taichi, then, they can get one when AMD launches UDNA in Q4 2026.

Like you said, there's no difference between the ones you mentioned, they're just aesthetically different is all.
 
@Sunny and 75 @eidairaman1

Thank you very much. What could be the reason for the asrock models being at the top of the 3dmark steelnomad tests? Could it be that this model is being given better silicon?

Also, we don't know if these values are stable in the game. But I wonder how much fps difference a 200-300 point difference in steelnomad would make in a 2k game? It would be great if someone used it.
 
@Sunny and 75 @eidairaman1

Thank you very much. What could be the reason for the asrock models being at the top of the 3dmark steelnomad tests? Could it be that this model is being given better silicon?

Also, we don't know if these values are stable in the game. But I wonder how much fps difference a 200-300 point difference in steelnomad would make in a 2k game? It would be great if someone used it.
Certain binnings and certain chips are going to be randomly better than others, depends on asic quality as well, which doesnt always mean higher OCs, keep in mind some people use Dry Ice, LN2, Waterblocks and Peltiers to cool the gpu dies
 
Certain binnings and certain chips are going to be randomly better than others, depends on asic quality as well, which doesnt always mean higher OCs, keep in mind some people use Dry Ice, LN2, Waterblocks and Peltiers to cool the gpu dies
Of course it is possible but my friend has a taichi model and with the settings attached he gets one point like this. There is no extra addition or operation on the computer.
 

Attachments

  • asdasd.png
    asdasd.png
    1.6 MB · Views: 35
Of course it is possible but my friend has a taichi model and with the settings attached he gets one point like this. There is no extra addition or operation on the computer.
My point is ymmv, and good for him. Get the card you desire and move forward.

If you want to spend more or less all the power to you, in the end it's your wallet feeling the effects.
 
I wonder how much fps difference a 200-300 point difference in steelnomad would make in a 2k game?
To know how a card performs while gaming, check out the related performance charts of the specific titles/types of games that you play.
 
@Sunny and 75 @eidairaman1

Thank you very much. What could be the reason for the asrock models being at the top of the 3dmark steelnomad tests? Could it be that this model is being given better silicon?

Also, we don't know if these values are stable in the game. But I wonder how much fps difference a 200-300 point difference in steelnomad would make in a 2k game? It would be great if someone used it.

Honestly they are all going to be margin of error different.

Screenshot 2025-05-30 221630.png

In TPU testing technically the best card was the Mercury but you could literally buy 10 of each and they would likely all trade blows.
 
He came on but never responded...
 
@Sunny and 75 @eidairaman1 @oxrufiioxo

Yes, nitro seems better at stock speeds. However, since I will be doing uv/oc + max power, I was wondering if a higher tdp taichi would make sense, and if it's actually worth it.

Best case scenario you get a gpu that's 2-4% better in synthetic workloads, worst case scenario you get a gpu that not only uses more power but clocks lower.

Silicon can vary by 1-2% +/- from chip to chip in my experiences so even if you got the worst Nitro possible and the best taichi possible we are talking 2-4% at best which isn't noticeable in normal gaming scenarios.

It's called the Silicon lottery after all AMD isn't sending better Silicon to a specific vendor but even if they were it would be sapphire or xfx their longest running partners that only make AMD dgpu.

Now if you plan is to do LN2 and break records on 3DMarks for a specific card you better buy 10 of each and keep the best one.

It really comes down to what card you like the aesthetics of the best becuase honestly how they look is what separates them the most beyond that just the noise profile of each but I'm guessing that's pretty similar.
 
Best case scenario you get a gpu that's 2-4% better in synthetic workloads, worst case scenario you get a gpu that not only uses more power but clocks lower.

Silicon can vary by 1-2% +/- from chip to chip in my experiences so even if you got the worst Nitro possible and the best taichi possible we are talking 2-4% at best which isn't noticeable in normal gaming scenarios.

It's called the Silicon lottery after all AMD isn't sending better Silicon to a specific vendor but even if they were it would be sapphire or xfx their longest running partners that only make AMD dgpu.

Now if you plan is to do LN2 and break records on 3DMarks for a specific card you better buy 10 of each and keep the best one.

It really comes down to what card you like the aesthetics of the best becuase honestly how they look is what separates them the most beyond that just the noise profile of each but I'm guessing that's pretty similar.
Thanks. Now I saw the news that 9080xt will be released. I am really confused. I am wondering if I should wait and get 9080xt instead of 9070xt.
 
Thanks. Now I saw the news that 9080xt will be released. I am really confused. I am wondering if I should wait and get 9080xt instead of 9070xt.

Assuming it's real don't expect less than 1200 usd or more.

I have my doubts it's going to be released anytime soon when it does it'll be significantly more than the 9070xt.
 
It doesn't matter. As far as I've seen, all 9070 XT models run at pretty much the same temperatures (provided you have adequate airflow). The rest is just a design choice, or thickness of the cooler.
 
@oxrufiioxo Frankly, the price is not a big problem. Even if it was $1200, I would still buy it. But the real question is when it will go on sale. If I use a 9070xt and sell it when a 9080 comes out, will I lose a lot?
 
Hard to say the gpu market is so volatile right now prices are inflated so if that card hits its msrp in the next 6 months you could lose 4-500 usd or prices could go up and you might lose nothing.

Nobody can predict where we will be at in 6-12 months anyone would just be guessing.
 
edit: I would be very surprised if we see a Radeon 9080. I highly doubt it will be on sale. AMD would not make a smart choice to produce such cards, especially as the 7000 high end cards and 6000 high end cards barely sold.

Personally I dislike when people buy more than one graphic card for themself when the demand is one graphic card.

The only valid reason for a return is, defective, fan defect, unbearable noise, damaged box. Item not as described and customer support can not answer questions, e.g. my recent noctua purchase for 100€ fans. Well partly also a scam from amazon.de as the seller was not the seller as shown. Noctua has two companies - a fraud at the end of the day. RASCOM is not Noctua.

I want to be the first customer who buys new hardware. Sellers seems to have sold me several times preused, 2 weeks or 6 weeks old hardware.
 
Personally I dislike when people buy more than one graphic card for themself when the demand is one graphic card.
Oh, I'm not gonna send you a picture of my PC shelf, then. :roll:

Sellers seems to have sold me several times preused, 2 weeks or 6 weeks old hardware.
That is illegal, at least by UK standards, as far as I know.
 
edit: I would be very surprised if we see a Radeon 9080. I highly doubt it will be on sale. AMD would not make a smart choice to produce such cards, especially as the 7000 high end cards and 6000 high end cards barely sold.

Personally I dislike when people buy more than one graphic card for themself when the demand is one graphic card.

The only valid reason for a return is, defective, fan defect, unbearable noise, damaged box. Item not as described and customer support can not answer questions, e.g. my recent noctua purchase for 100€ fans. Well partly also a scam from amazon.de as the seller was not the seller as shown. Noctua has two companies - a fraud at the end of the day. RASCOM is not Noctua.

I want to be the first customer who buys new hardware. Sellers seems to have sold me several times preused, 2 weeks or 6 weeks old hardware.

I honestly think it would sell if they releases it today as long as it on average bests the 5080 and at least got close to the 4090 on average. The problem is the investment cost and what the market is going to look like in 6-12 months those resources should likely be poured into the next generation unless they feel this is going to be another long generation and they want a refresh late this year or early next.

5080 is 23% faster in tpu latest gpu review at 4k so if this can overtake that I could see an msrp of 1000 and a street price of 1200 the 5090 would still cost double so that's too large a jump to worry about that card 1000 usd 7900XTX wasn't as appealing becuase it was only good at 1 thing raster and it was a generation behind at everything else also both the 4080 and 4090 weren't hard to get at msrp unlike 2025 GPUs at least for the first 5-6 months in the 4090s case.
 
Back
Top