• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

AMD Navi 23 ‘NVIDIA Killer’ GPU Rumored to Support Hardware Ray Tracing, Coming Next Year

Status
Not open for further replies.
Navi was on the roadmaps in Raja's time. If you look at the Navi timeline,he left in late 2017. Yet in 1H 2018,there was leaks about Navi test samples being already out. There even flags pointing to Navi in drivers in late 2017. So unless people think a brand new GPU uarch,the 1st new one for AMD since GCN,was designed from scratch and had test samples in like under 12 months,then I have a bridge to sell you. The project was initiated during his tenure,at least at an early stage.

Think how long it took to design Ryzen. The design process started in 2015:
https://www.pcworld.com/article/3111388/inside-amds-development-of-the-zen-cpu.html

So nearly 2 years.
 
Its easy to pin it on a scapegoat. I dont think Intel would hire him if he was useless. I dont know the guy so just saying.. :)
 
Th0nt said:
Its easy to pin it on a scapegoat. I dont think Intel would hire him if he was useless. I dont know the guy so just saying.. :)
Click to expand...

Intel hired Murthy Renduchintala and it looks like Jim Keller may have left because of his politicking, Murthy eventually "left Intel too" executives are never sacked, they always leave or retire, doesn't mean it was a choice.
 
CAT-THE-FIFTH said:
Navi was on the roadmaps in Raja's time. If you look at the Navi timeline,he left in late 2017. Yet in 1H 2018,there was leaks about Navi test samples being already out. There even flags pointing to Navi in drivers in late 2017. So unless people think a brand new GPU uarch,the 1st new one for AMD since GCN,was designed from scratch and had test samples in like under 12 months,then I have a bridge to sell you. The project was initiated during his tenure,at least at an early stage.

Think how long it took to design Ryzen. The design process started in 2015:
https://www.pcworld.com/article/3111388/inside-amds-development-of-the-zen-cpu.html

So nearly 2 years.
Click to expand...

Agreed, but also a lot of time to improve, refine and iron-out problems, change aspects of the micro-architecture etc. He can't claim credit for the final product, but hey we can all roast him if the base design is flawed beyond redemption I guess. Best of both worlds.
 
humbug said:
Quality vs Quantity, Intel have about 120,000 staff, what are they all doing? Sipping coffee.... AMD have about 11,000 and i bet every single one of them has their specially and work to do.
Click to expand...


Well the quality certainly wasn't prevalent when Vega got kicked out the door, which people always put the blame on Koduri for when realistically it was budget and lack of dedicated engineers on the project. Vega was always a dual purpose gpu and again that comes down to budget and working with what they had available. It's still a decent gpu even today, just wasn't the world beater people expected.
 
humbug said:
Raja left / Was sacked whatever... Super Sue walked into the GPU division, said "hold my bear...." and increased Vaga IPC by 50%, BOOM, mic drop. Bye Raja :p
Click to expand...

Vega benefited from a die shrink enabling higher clocks, which was always going to happen for the pro series cards. It was another guy at amd who allegedly pushed for it to be made available as a 1080ti competitor. Can't recall who but there was a story on it a few months before radeon vii was announced.
 
Th0nt said:
Its easy to pin it on a scapegoat. I dont think Intel would hire him if he was useless. I dont know the guy so just saying.. :)
Click to expand...
ik9000 said:
Agreed, but also a lot of time to improve, refine and iron-out problems, change aspects of the micro-architecture etc. He can't claim credit for the final product, but hey we can all roast him if the base design is flawed beyond redemption I guess. Best of both worlds.
Click to expand...

I think some of the blame on him was unfair though. AMD R and D had been cut during Rory Read's time as if they hadn't reduced costs they would have gone bankrupt. Then they diverted a larger percentage of the R and D towards Zen development. On top of this AMD for the first time due to WSA,used GF for its GPUs instead of only its CPUs.

You can see the rot set in after 2014,when didn't have the money to tackle Maxwell. All we saw was relatively minor changes here and there to the basic GCN uarch. For instance its taken well over 5 years to move past the 64CU limit the original GCN GPUs had.

This is the same GF,which scrapped its own internally developed 20NM and 14NM process nodes,and had to license a Samsung 14NM process originally made for small mobile GPUs. Nvidia only used it for sub 150MM2 GPUs. Vega was probably the biggest GPU made on Samsung/GF 14NM. Much of those horrible power requirements are probably the process node - even Ryzen power consumption went wonky if the CPUs were pushed and overvolted too much.

Just look at Polaris,AMD had working samples months before launch. Yet,look at how Polaris launched?? It was clear the PCB power connectors,cooler,etc were all made for a GPU which consumed less power than what we got. Its most likely GF ***** up yet again,and AMD had to overvolt the GPUs,to get performance upto what they originally expected! Remember all the rumours about Zen,not clocking high enough,which lead to motherboard OEMs underspeccing motherboards?? Well Zen IIRC,was meant to be late 2016. Yet it launched well into 2017.
 
Gerard said:
Vega benefited from a die shrink enabling higher clocks, which was always going to happen for the pro series cards. It was another guy at amd who allegedly pushed for it to be made available as a 1080ti competitor. Can't recall who but there was a story on it a few months before radeon vii was announced.
Click to expand...
Vega 8 is 73% faster overall, its not just clocks, AMD themselves said they increased Vega IPC by about 50%.
 
Gerard said:
Talking about radeon vii, vega 8 is for notebooks.
Click to expand...

RDNA1 is 50% faster per CU per clock, and this with half the memory bandwidth, despite this it is a 1080TI competitor, so is the 5700XT

Yes it benefited from a die shrink, its still crap.
 
Status
Not open for further replies.
Back
Top Bottom