Caveman Tech @Cavemtech

https://t.co/XsEzJCkV5n Joined August 2021

Tweets

1K
Followers

1K
Following

199
Likes

3K

Caveman Tech @Cavemtech

23 hours ago

@mikadontlouz La fem Nikita

0 0 0 12 0

View Details

Caveman Tech @Cavemtech

3 days ago

@Liquid_BTC @Blockstream

0 0 0 65 0

View Details

Caveman Tech @Cavemtech

4 days ago

What color should we dye it @we_aint_worried ?

1 0 0 71 0

View Details

I want to show you how wild AI is right now. One prompt to Claude: "make a Clash of Clans style game." 30 min: playable. Same chat, one day: 3D, AI-rendered art and creatures, animation, music, multiplayer, critters that level up. I never wrote a line of code. I just described what I wanted.

2 0 3 506 1

View Details

Caveman Tech @Cavemtech

5 days ago

@RedPandaMining

0 0 1 233 0

View Details

Caveman Tech @Cavemtech

a week ago

Claude Fable 5.1 is out. Any initials verdicts ?

0 0 0 72 0

View Details

Caveman Tech @Cavemtech

a week ago

@RedPandaMining Bro same here

0 0 0 123 0

View Details

Caveman Tech @Cavemtech

a week ago

@HawkCryptoTech If the apple and cheese trio was included than that’s a fucking steal 🤣

1 0 1 61 0

View Details

Caveman Tech @Cavemtech

a week ago

Fable 5 promo is over. We all got nerfed. $100 a month Max plan is wayyy less attractive now. I don’t know about yall but I’m hitting limits so fast now. #claude #fable5 #ai

1 0 2 152 0

View Details

Caveman Tech @Cavemtech

2 weeks ago

@bokiko I miss those guys

1 0 1 40 0

View Details

Caveman Tech @Cavemtech

2 weeks ago

@RedPandaMining gotta prime them bad boys for max useage :)

0 0 1 98 0

View Details

Red Panda Mining @RedPandaMining

2 weeks ago

NO WAY…

29 7 187 12K 11

View Details

Caveman Tech @Cavemtech

2 weeks ago

@RedPandaMining

1 0 1 77 0

View Details

Caveman Tech @Cavemtech

2 weeks ago

@RedPandaMining Makes me wanna

0 0 0 186 0

View Details

Caveman Tech @Cavemtech

2 weeks ago

@SonOfATech That's the design. X placed us "users" in a jar like spiders and shakes it to make us attack each other. Politics gets dragged into everything — even an operating system — because political outrage is the hardest shake. And X profits off every fight.

0 0 0 33 0

View Details

Caveman Tech @Cavemtech

2 weeks ago

@RedPandaMining How much ????

0 0 0 30 0

View Details

Caveman Tech @Cavemtech

3 weeks ago

Benchmark is done via Pytorch Steps here: python import torch # Ensure you have a CUDA device available device = "cuda" if torch.cuda.is_available() else "cpu" if device == "cpu": raise SystemError("A CUDA GPU is required to accurately measure TFLOPS.") # 1. Choose a large matrix size to fully saturate the GPU N = 4096 A = torch.randn(N, N, dtype=torch.float32, device=device) B = torch.randn(N, N, dtype=torch.float32, device=device) # 2. Turn OFF TF32 to force true, strict FP32 precision torch.backends.cuda.matmul.allow_tf32 = False # 3. Warm up the GPU (prevents lazy initialization overhead) for _ in range(10): torch.matmul(A, B) torch.cuda.synchronize() # 4. Measure execution time across multiple iterations start_event = torch.cuda.Event(enable_timing=True) end_event = torch.cuda.Event(enable_timing=True) iterations = 50 start_event.record() for _ in range(iterations): torch.matmul(A, B) end_event.record() # Wait for GPU to finish operations torch.cuda.synchronize() # 5. Calculate TFLOPS elapsed_time_ms = start_event.elapsed_time(end_event) avg_time_seconds = (elapsed_time_ms / 1000) / iterations # FLOPs formula for matmul: 2 * N^3 total_flops = 2 * (N ** 3) achieved_tflops = (total_flops / avg_time_seconds) / 1e12 print(f"Matrix Size: {N}x{N}") print(f"Average Execution Time: {avg_time_seconds*1000:.3f} ms") print(f"Achieved FP32 Performance: {achieved_tflops:.2f} TFLOPS")